From 5f9efbba56a5afc710caf9a34f97975974f2d380 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Tue, 9 Jun 2026 14:07:09 -0700 Subject: [PATCH 01/12] draft of data management chapter --- 03-data_management.Rmd | 139 ++++++++++++++++++++++++++++++++++++++++- 1 file changed, 138 insertions(+), 1 deletion(-) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index 011f3e2..8bac686 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -12,10 +12,147 @@ ottrpal::set_knitr_image_path() ## Clinical data handling tools +Clinical research relies heavily on the collection, processing, analysis, and management of data. Efficient and effective data handling is crucial to ensure the integrity, reliability, and validity of clinical trials and studies. Selecting the right tools for handling clinical data is a critical step in the research process. This section defines data handling in the context of clinical research and contrasts it with data quality, discusses the importance of privacy, provides a comprehensive guide on how to find suitable clinical data handling tools with a focus on open-source options, and emphasizes the importance of testing data handling methods using dummy data in the preliminary phases of a clinical trial. + +### Clinical data handling + +**Clinical data handling** involves the processes and tools used to collect, manage, store, process, and share clinical data in a way that maintains its integrity, security, and usability. It encompasses a range of activities, including: + +- Data Collection: Gathering data from various sources such as case report forms (CRFs), electronic health records (EHRs), patient surveys, and laboratory results. +- Data Storage and Management: Safely storing data in databases or data warehouses and managing access to ensure that only authorized personnel can interact with the data. +- Data Cleaning and Processing: Ensuring data consistency, accuracy, and completeness by detecting and rectifying errors, missing values, and inconsistencies. +- Data Analysis: Using statistical and analytical tools to generate insights from the data that can inform clinical decisions and study outcomes. +- Data Sharing and Reporting: Providing access to data for collaborators, regulators, or stakeholders in a secure and controlled manner and generating reports that summarize findings. + +The goal of effective data handling is to ensure that data is accurate, reliable, and available when needed, while also protecting patient privacy and complying with regulatory standards. + +#### Data quality vs data handling + +Data quality and data handling are closely related but distinct concepts in clinical research. + +- **Data Quality**: Refers to the accuracy, completeness, consistency, and reliability of data. High-quality data is essential for producing valid and reliable research outcomes. It involves processes such as data validation, error checking, and quality control measures. +- **Data Handling**: Encompasses the broader scope of managing the data lifecycle, from collection and storage to processing, analysis, and sharing. While data quality is a component of data handling, the latter also involves aspects like data security, access management, and regulatory compliance. + +Both data quality and data handling are critical for ensuring the integrity of clinical research. Data handling tools must support high data quality standards through features like automated data validation checks, error reporting, and data cleaning functions. + +### Privacy considerations + +Privacy is a fundamental consideration in clinical data handling, given the sensitive nature of the information involved. Clinical data often contains personally identifiable information (PII) or personal health information (PHI), which must be protected to comply with privacy regulations such as the Health Insurance Portability and Accountability Act (HIPAA) in the United States, the General Data Protection Regulation (GDPR) in the European Union, and other regional laws. + +- **Data Anonymization and De-identification**: One of the primary methods to protect privacy is to anonymize or de-identify data, removing or encrypting any information that could directly or indirectly identify an individual. +- **Privacy-Preserving Record Linkages**: In scenarios where data from multiple sources need to be linked without compromising individual privacy, privacy-preserving record linkage techniques are essential. These methods enable the integration of datasets by matching records in a way that minimizes the risk of re-identification. Techniques such as secure multi-party computation, homomorphic encryption, and differential privacy can be employed to ensure that the linkage process itself does not expose sensitive information. +- **Access Control and Encryption**: Tools should support robust access controls, ensuring that only authorized users have access to the data. Encryption should be used for both data at rest and data in transit to prevent unauthorized access. +- **Compliance and Auditing**: Tools should facilitate compliance with regulatory standards and provide auditing capabilities to track data access and usage. + +When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research. + +### Selecting clinical data handling tools + +Choosing the right tools for clinical data handling depends on several factors, including the study's size, complexity, data types, regulatory requirements, and budget. Key considerations include: + +- **Compatibility and Integration**: The tool should be compatible with existing data systems and workflows, allowing for seamless integration of data from multiple sources. +- **Scalability**: The tool should be able to handle the volume of data expected in the study and scale as the study progresses. +- **User-Friendliness**: A user-friendly interface can reduce the learning curve for researchers and data managers, improving efficiency and reducing errors. +- **Regulatory Compliance**: The tool should comply with relevant data privacy and security regulations to protect patient data. +- **Support and Documentation**: Adequate support, including user manuals, tutorials, and customer service, is essential for troubleshooting and maximizing the tool's potential. + +Research teams should conduct thorough evaluations, including reading reviews, seeking recommendations, and conducting pilot tests, to identify the best tool for their specific needs. + +#### Open source options + +Open-source tools provide a cost-effective and flexible alternative to proprietary software for handling clinical data. These tools are often developed and maintained by vibrant communities and can be customized to fit specific research needs. + +Advantages of open-source tools include: + +- They are typically **free to use** +- They provide **transparency** by allowing users to review the codebase for security and compliance +- They can be **modified** to meet unique requirements +- They often have strong **community support**, which can be valuable for troubleshooting and enhancing functionalities. + +Examples of popular open-source data handling tools include: + +- **REDCap**: A secure, web-based application for building and managing online surveys and databases, widely used in clinical research for data collection and management. +- **OpenClinica**: An open-source clinical trial software that supports data capture, management, and analysis, compliant with regulatory standards like FDA requirements for electronic data (FDA 21 CFR Part 11). +- **KNIME**: A data analytics, reporting, and integration platform that can be used for data cleaning, processing, and analysis, with extensive capabilities for machine learning and data visualization. +- **R and Python**: These programming languages offer powerful libraries and packages (such as the `tidyverse` for R, and `pandas` and `NumPy` for Python) that support a wide range of data handling and analysis tasks. + +Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform that offers some basic statistical and data visualization tools, and offers APIs to allow the real-time analytics in other platforms (R, C#, cURL, etc.). However, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued [guidance](https://www.r-project.org/doc/R-FDA.pdf) on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR. + +While open-source tools are beneficial, it is essential to ensure they are secure, well-maintained, and compliant with relevant privacy and regulatory standards. + +When moving between multiple data handling tools, it is important to consider interoperability. As an example, REDCap and R play very nicely together via API tokens, but this may not be true of all electronic data capture systems or databases. The sharing, integration and redistribution of data between systems - whether internally or externally - needs to be carefully planned, tested and documented. + +### Testing data handling tools + +Before implementing data handling methods in an actual clinical trial, it is crucial to run tests using **dummy data**. This step ensures that the data handling process is robust, efficient, and free from errors across different conditions, without risking sensitive patient information. + +- **Creating Realistic Dummy Data**: Dummy data should mimic the real data in terms of format, structure, and complexity. It should include various scenarios (e.g., missing data, outliers, data entry errors) to test the system's error-handling capabilities. +- **Simulating the Full Data Handling Workflow**: The process should involve every stage of data handling, from data collection and entry to storage, cleaning, analysis, and reporting. This comprehensive simulation helps identify potential issues early, such as data loss, security vulnerabilities, or errors in data processing algorithms. +- **Refining and Optimizing Data Handling Methods**: Based on the findings from dummy data tests, researchers can refine their data handling protocols, adjust tools and settings, and optimize workflows to ensure smooth operations when real data is introduced. + +Testing with dummy data provides an additional layer of quality assurance and helps build confidence in the data handling process before the clinical trial begins. + +### Conclusion + +Effective clinical data handling is fundamental to the success of clinical trials and studies. Selecting the right tools requires careful consideration of privacy, regulatory compliance, open-source versus proprietary options, and the specific needs of the study. Understanding the difference between data quality and data handling is crucial, as both are essential for ensuring the validity and reliability of research outcomes. Running preliminary tests using dummy data is a critical step in validating data handling methods, ensuring that the chosen tools and processes are robust, secure, and efficient. By following these guidelines, researchers can enhance data management practices, protect patient privacy, and achieve meaningful and reliable clinical research outcomes. + ## Importance of keeping clinical data safe ## Government regulators +Clinical data is governed by several different types of regulations. In this section, we will review some of the major regulatory frameworks and organizations. + +### Health Insurance Portability and Accountability Act (HIPAA) + +The [Health Insurance Portability and Accountability Act](https://www.hhs.gov/hipaa/for-professionals/privacy/laws-regulations/index.html) (HIPAA), regulated by the U.S. Department of Health and Human Services (HHS), establishes national standards to protect individuals’ medical records and other personal health information. It applies to health plans, healthcare clearinghouses, and healthcare providers that conduct certain healthcare transactions electronically. The HIPAA Privacy Rule requires appropriate safeguards to protect the privacy of protected health information (PHI) and sets limits on the uses and disclosures of such information without patient authorization. + +### Food and Drug Administration (FDA) + +The U.S. [Food and Drug Administration](https://www.fda.gov/regulatory-information/fda-rules-and-regulations) (FDA) regulates the safety, efficacy, and security of human and veterinary drugs, biological products, medical devices, food, cosmetics, and products that emit radiation. The FDA’s regulations ensure that clinical trials are conducted ethically and that data collected is reliable and accurate. This includes oversight of clinical trial protocols, informed consent, and reporting of adverse events. + +### General Data Protection Regulation (GDPR) + +The [General Data Protection Regulation](https://gdpr.eu/what-is-gdpr/) (GDPR), governed by the European Union (EU), is a comprehensive data protection law that governs the collection, processing, storage, and transfer of personal data within the EU. It aims to enhance individuals’ control over their personal data and simplify the regulatory environment for international business. GDPR applies to any organization that processes the personal data of EU residents, regardless of where the organization is based. Key provisions include the right to be forgotten, data portability, and mandatory breach notifications. + +### Honest brokers + +Honest brokers act as neutral intermediaries between the data source and researchers, typically regulated by Institutional Review Boards (IRBs) or equivalent bodies. They are responsible for de-identifying data to ensure that researchers cannot trace the data back to individual patients. Honest brokers must complete specific training, such as Collaborative Institutional Training Initiative (CITI) Research Ethics and HIPAA training, before accessing data. They play a crucial role in maintaining the confidentiality and integrity of clinical data. + ## Documentation -## Conclusion \ No newline at end of file +In data management, several types of documentation are frequently encountered, each serving a specific purpose throughout the lifecycle of a research project. These documents are crucial for ensuring data integrity, regulatory compliance, and effective project management. + +### Data Management Plan + +A Data Management Plan (DMP) outlines how data will be handled during and after a research project. Typically created at the beginning of a project, it is stored in project documentation repositories or institutional databases. Often required by funding agencies, a DMP may include plans for testing with dummy data to ensure data integrity. The importance of a DMP lies in its role in planning and managing data throughout the project, ensuring that data is handled consistently and [responsibly](https://www.tableau.com/learn/articles/what-is-data-management). + +### Statistical Analysis Plan + +The Statistical Analysis Plan (SAP) details the statistical methods and analyses to be performed on the collected data. Developed before data analysis begins, it is stored with project documentation or in electronic lab notebooks. Essential for clinical trials and other research requiring rigorous statistical analysis, the SAP may involve testing statistical methods on dummy data. The SAP is crucial for maintaining the integrity and reproducibility of statistical [analyses](https://www.questionpro.com/blog/data-documentation/). + +### Standard Operating Procedures + +Standard Operating Procedures (SOPs) provide detailed instructions on how to perform specific tasks or processes. Used throughout the project lifecycle, they are stored in organizational repositories or document management systems. SOPs are often required for regulatory compliance and may include procedures for testing with dummy data. The importance of SOPs lies in their ability to standardize processes, ensuring consistency and compliance with regulatory [requirements](https://managing-qualitative-data.org/modules/2/a/). + +### Data Use Agreements + +Data Use Agreements (DUAs) define the terms and conditions for data sharing and use. Encountered before data is shared with external parties, they are stored in legal or administrative offices. DUAs are required when sharing data with external collaborators but are typically not directly related to dummy data. DUAs are important for protecting data privacy and ensuring that data is used [appropriately](https://www.ibm.com/topics/data-compliance). + +### Data Sharing Agreements + +Data Sharing Agreements (DSAs) specify the terms for sharing data between organizations. Encountered prior to data sharing, they are stored in legal or administrative offices. DSAs are necessary for formalizing data sharing arrangements and generally do not relate to dummy data. The importance of DSAs lies in their role in facilitating collaboration while protecting data integrity and compliance with legal [requirements](https://atlan.com/know/data-governance/data-compliance-management/). + +### Documentation across the span of a project + +These documents are encountered at various stages of a project, from planning (DMP, SAP) to execution (SOP) and data sharing (DUA, DSA). They are typically stored in project documentation repositories, institutional databases, or document management systems, with access restricted to authorized personnel. Not all documents are required for every project; their necessity depends on the project’s scope, regulatory requirements, and institutional policies. Some documents, like the DMP, SAP, and SOP, may include provisions for testing with dummy data to ensure data integrity and validate processes. + +To ensure compliance with these documents, organizations should implement robust data governance frameworks that include regular audits, training programs, and clear policies and [procedures](https://teamhub.com/blog/document-management-compliance-2/). Standardized templates for these documents are often available from funding agencies, regulatory bodies, or institutional guidelines, helping to ensure consistency and compliance with best practices. It’s best to ask before starting any document to ensure the correct format is used. + +For further reading, consider the following resources: + +- Guide to Clinical Data Management Procedures (GCDMP): GCDMP +- Books by Suzanne Prokscha: Writing and Managing SOPs for GCP and Practical Guide to Clinical Data Management + +Additionally, remember the insightful quote by [Damian Conway](https://www.questionpro.com/blog/data-documentation/): “Documentation is a love letter that you write to your future self.” This is an invaluable tidbit to keep in mind throughout the lifecycle of a study. + +## Conclusion From bd0c71519f6c858f793f0255a6022913643a0e7f Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 10:02:53 -0700 Subject: [PATCH 02/12] move privacy section, add learning objectives --- 03-data_management.Rmd | 26 ++++++++++++++------------ 1 file changed, 14 insertions(+), 12 deletions(-) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index 8bac686..2f11adc 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -10,6 +10,10 @@ ottrpal::set_knitr_image_path() ## Learning Objectives +```{r, fig.align='center', echo = FALSE, fig.alt= "Learning Objectives: Learning Objectives: 1. Understand the clinical data handling process and how to choose clinical data handling tools, 2. Explain privacy considerations for clinical data, 3. List major regulations and regulatory bodies for clinical data, 4. Describe different types of documents used when working with clinical data", out.width="100%"} +ottrpal::include_slide("https://docs.google.com/presentation/d/1ivDTcLjb2078O0GemkSeCgC1jmxk4fMsiFQaPaer9mQ/edit?slide=id.g3edc229d226_1_0#slide=id.g3edc229d226_1_0" ) +``` + ## Clinical data handling tools Clinical research relies heavily on the collection, processing, analysis, and management of data. Efficient and effective data handling is crucial to ensure the integrity, reliability, and validity of clinical trials and studies. Selecting the right tools for handling clinical data is a critical step in the research process. This section defines data handling in the context of clinical research and contrasts it with data quality, discusses the importance of privacy, provides a comprehensive guide on how to find suitable clinical data handling tools with a focus on open-source options, and emphasizes the importance of testing data handling methods using dummy data in the preliminary phases of a clinical trial. @@ -35,17 +39,6 @@ Data quality and data handling are closely related but distinct concepts in clin Both data quality and data handling are critical for ensuring the integrity of clinical research. Data handling tools must support high data quality standards through features like automated data validation checks, error reporting, and data cleaning functions. -### Privacy considerations - -Privacy is a fundamental consideration in clinical data handling, given the sensitive nature of the information involved. Clinical data often contains personally identifiable information (PII) or personal health information (PHI), which must be protected to comply with privacy regulations such as the Health Insurance Portability and Accountability Act (HIPAA) in the United States, the General Data Protection Regulation (GDPR) in the European Union, and other regional laws. - -- **Data Anonymization and De-identification**: One of the primary methods to protect privacy is to anonymize or de-identify data, removing or encrypting any information that could directly or indirectly identify an individual. -- **Privacy-Preserving Record Linkages**: In scenarios where data from multiple sources need to be linked without compromising individual privacy, privacy-preserving record linkage techniques are essential. These methods enable the integration of datasets by matching records in a way that minimizes the risk of re-identification. Techniques such as secure multi-party computation, homomorphic encryption, and differential privacy can be employed to ensure that the linkage process itself does not expose sensitive information. -- **Access Control and Encryption**: Tools should support robust access controls, ensuring that only authorized users have access to the data. Encryption should be used for both data at rest and data in transit to prevent unauthorized access. -- **Compliance and Auditing**: Tools should facilitate compliance with regulatory standards and provide auditing capabilities to track data access and usage. - -When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research. - ### Selecting clinical data handling tools Choosing the right tools for clinical data handling depends on several factors, including the study's size, complexity, data types, regulatory requirements, and budget. Key considerations include: @@ -96,7 +89,16 @@ Testing with dummy data provides an additional layer of quality assurance and he Effective clinical data handling is fundamental to the success of clinical trials and studies. Selecting the right tools requires careful consideration of privacy, regulatory compliance, open-source versus proprietary options, and the specific needs of the study. Understanding the difference between data quality and data handling is crucial, as both are essential for ensuring the validity and reliability of research outcomes. Running preliminary tests using dummy data is a critical step in validating data handling methods, ensuring that the chosen tools and processes are robust, secure, and efficient. By following these guidelines, researchers can enhance data management practices, protect patient privacy, and achieve meaningful and reliable clinical research outcomes. -## Importance of keeping clinical data safe +## Privacy considerations for clinical data + +Privacy is a fundamental consideration in clinical data management, given the sensitive nature of the information involved. Clinical data often contains personally identifiable information (PII) or personal health information (PHI), which must be protected to comply with privacy regulations such as the [Health Insurance Portability and Accountability Act](https://www.hhs.gov/hipaa/for-professionals/privacy/laws-regulations/index.html) (HIPAA) in the United States, the [General Data Protection Regulation](https://gdpr.eu/what-is-gdpr/) (GDPR) in the European Union, and other regional laws. More information about PII and PHI can be found in [this course](https://hutchdatascience.org/Ethical_Data_Handling_for_Cancer_Research/data-privacy.html) about ethical data handling. + +- **Data Anonymization and De-identification**: One of the primary methods to protect privacy is to anonymize or de-identify data, removing or encrypting any information that could directly or indirectly identify an individual. +- **Privacy-Preserving Record Linkages**: In scenarios where data from multiple sources need to be linked without compromising individual privacy, privacy-preserving record linkage techniques are essential. These methods enable the integration of datasets by matching records in a way that minimizes the risk of re-identification. Techniques such as secure multi-party computation, homomorphic encryption, and differential privacy can be employed to ensure that the linkage process itself does not expose sensitive information. +- **Access Control and Encryption**: Tools should support robust access controls, ensuring that only authorized users have access to the data. Encryption should be used for both data at rest and data in transit to prevent unauthorized access. +- **Compliance and Auditing**: Tools should facilitate compliance with regulatory standards and provide auditing capabilities to track data access and usage. + +When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research. ## Government regulators From d302733fc8e620055fd06779e71b33678f334700 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 10:18:42 -0700 Subject: [PATCH 03/12] updates to docs etc when working on ch 5 --- .gitignore | 10 - Course_Name.rds | Bin 257 -> 327 bytes .../figure-html/unnamed-chunk-2-1.png | Bin 96493 -> 0 bytes .../figure-html/unnamed-chunk-3-1.png | Bin 66074 -> 0 bytes docs/03-data_management.md | 139 ++++++++++++- ...C1jmxk4fMsiFQaPaer9mQ_g3edc229d226_1_0.png | Bin 0 -> 66775 bytes docs/404.html | 60 +++--- docs/clinical-data-management.html | 195 ++++++++++++++---- docs/reference-keys.txt | 21 ++ docs/search_index.json | 2 +- 10 files changed, 349 insertions(+), 78 deletions(-) delete mode 100644 .gitignore delete mode 100644 _bookdown_files/02-chapter_of_course_files/figure-html/unnamed-chunk-2-1.png delete mode 100644 _bookdown_files/02-chapter_of_course_files/figure-html/unnamed-chunk-3-1.png create mode 100644 docs/03-data_management_files/figure-html/1ivDTcLjb2078O0GemkSeCgC1jmxk4fMsiFQaPaer9mQ_g3edc229d226_1_0.png diff --git a/.gitignore b/.gitignore deleted file mode 100644 index dc638be..0000000 --- a/.gitignore +++ /dev/null @@ -1,10 +0,0 @@ -# ignore nonsense files -.DS_Store -.Rhistory -.local -.rstudio -.bash_history -spell_check_results.tsv -.RData -.httr-oauth -docker/git_token.txt diff --git a/Course_Name.rds b/Course_Name.rds index 89313df82ae15a4ab55e98502667db77425f4979..03b31e9deba3779ae4932df1147ebda6d58a013f 100644 GIT binary patch literal 327 zcmV-N0l5AjiwFP!000001C>%uPlG@Zo$`?sYc0lTjQ$49Ry}z2(!?v##2eYdU^dE@ zz&5FWzI9nzmW4``%kI4QW@edr? z+}vdefMd0A?~^kTWhkf&szm4tQuc*ql(0eu?TQmsF;a9VXUe$1BEy7HQIRUI`nKH~ zgalGstklB{$|v(6p@K%@JIAum-;nt-D07Ti%Hn89IV1OVrFFTdjmcb+hnmhPryrOh z7tNSPJ98@5u2cKEpWL2^eb68K`5Ac(Xr3ca*yg4GuJQA7wf@IbCQ3il-6%f|F~tIT zyr0RRdiXkNsF7g?m(#26KZO2e{u7T6W2^u>$Eue|p-*&GhzOi(TUme+YJFC4C ZjeGSjG;r3MUZ@olqQ8>`;SCvByP(&qWdebgc&N zUIkYyK^0JqV!_kjtL4gDu6*j1X=SXF zfBUfVmh9+5X)4?KACc*D?T44(PL^02voL+g49x}Szv#x|z3n-DD04Nh=rN H>;eD)&&qvj diff --git a/_bookdown_files/02-chapter_of_course_files/figure-html/unnamed-chunk-2-1.png b/_bookdown_files/02-chapter_of_course_files/figure-html/unnamed-chunk-2-1.png deleted file mode 100644 index 0a0abd239f061342ac03e6c68c3ddc09ad7c8dcd..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 96493 zcmeFZWmuG3*f0u+qNpH<(t;w=paLR2N-3>$qbMEH(y|djQlyj|T3WgpR7#N=O1fj{ zj)9puYe4sgz4_iB-@kLt`-AIxnDsm>?yhyeRg{-HPe?_GgM)Ma;REr7)2NV% zM3)t?U2pN~Oig`Hgysz^G=JRwx^D*kyuf0BJ=DDL-n;MfgRv$zxX1UBO;|+8?p^ru zX(V(6qxFU9s%hqeKSyGrVj{j5^L86jDgjP@X0D98pdz%?tg2~XXDVjNeR{}tap>&3 zQg*F~Ax1oQsvM5VjrEiBR35A|%8Y+;P% zaulU-eFwdNW!?R8ZR|M?`chrK&MW+TyNzdho^GG16*P~bA0z!%gri0gN4%Wh`Z46T zubu=ub4CEVw%?avnqgBDpTKsxKS+r8Q#Q-BCW-HNGNa@fI;y;OJ&Gj;&1(C-6oLzG zx{(c^DBX-?wTT$@xe%{1F<|GL%V|*BGA*upJM2DkA2a9e3e8sO{7*WJudP-N{^$2te#Q$jc&|2 zyT+ULdmBgxwb+lR$KLakpQQ?5raR>YKiL9f`(BY13}QLaDq2W~tpo@!Dt zlrD5ZTcq~VhrsWUFRz{EFWuC?DE^L~*u+&!KPionYQ{KWCdD{myK7{A_$$i}8mC0r z6x}#C(|+@QDu!XXE#vEwu?j=wId>=HO#8MohA$X8Z$0FC_}*)q=33~)P0FiD^zWIz zD?JN%)-E7T#M=;Y&Gy+^S$n4wF2mz^KL1Mp8OM9|XrpHzMhaW&q~rN6{A&HxYW5?% z$5L0o%~H+3LQ*xpqT3c>cio7}l*6vEuQS%B74ubZy63arG@s@-E#@ZW5l)#zJ{pL>ulb8Zi20ef z@Yio&pUEi9Tzd2R%o8qW2foaUf#fj^`rTf zP8&pi6VX_JNo-+J)U4mP?-S!s33G3=UOqQj17)5opQ)f}PbDgfG@iuy>~j0=SF49; z6nVc>%5&TJQ+M~61h+Jc`H4rePPFYBH{E$Y+a`W*^rrM{&cz3(JzI{QOX~{T*^UJkL7qg3Uc3V5qIgC>Z#hkz}zC2rJ(A@+C#NFrW4-Mtya9F|f zX&hW16C8Z-gbV(tz#k3{9u)7-FDD_;6Mvph)L;jTNxnV8!GYjB6u+nZ5_f*^OpLb5 z$D4~K86|eA^z8Ys<=)}s=l_^L!}DAD}3Uk+VHRW z_E&g^vq5uYmEY?^pPal7p>{B~4>)eMGm5SM-QWK^meQmyHsW%e4-#DT>pyeQ~9>dAXmr3g_(@7 ziE5Jnb3mfWqB0sYZBzDuFQp70iNS2q80X~Wt{1P;EF<+uOyXeSJ(x}d3?|xWV0Wt! z{EC6Wq-_Us1Kd(dk~U~Qh`2e^yr{tqH$^c*pz;2$xaHNe5+>av);BPL9k(hX0g0Lyl2N+C$f4@O}0A+KRbyZ-UKY7DnC(r&M zHM#w*GhHd2=OY9eRPyZmdv&8&`>3qx^8TKl67#>5orxa|Z%ApvNzzJ&oaY{7<_Y8C zjnlWii{DmyI&P-7tCm1IBnDlWdGS8aw%UH2<4 zE^cOVaZ1!LRU@615yDnbt~8cpBg(eW(-LyFu&^+!K{&8#g|5t~+jK^B3Q>)3+%5TH zVnJ>^u6l3NgEOE~mG{{_dPWHAcpO@F(I{b|@2&uz$KN(L^C^zFx0(kcW+vhF7$>RRX0WM2OPNqZ+HNS_c2WAIe(gG;9T2T z&Qkit%C(leW4;Ve=dw$Jb{8wG)RP`Pe3e=|o!4-M8ezJMAWyRU^pmqOj*tLc+@&fpxw0@$B5AEY@TlgnKtrX&foS+w)>ecCK@xm?l z8Ys$%G8KG;RSaE|G5+CncV|DRNAXDFg-=7uaqdS==JG^B1SYa=D6|FCUWyFANE5fp z!KIb7>TrP|Vey=QC4lj$jc|F7_tNDQCgm*6jPL42g^67WjIl`0c16o&|$1=#}FsYyR$SV`yt`k1uX)3 z&3F34h`khiYKbFofCScI7(B(wq7h9BRZ5k8OUZ3`j_heK#o1&rJte#_424+FlMR&U zUNTor+zI=;4-5v9*Xx)VA#k(LnKePT9B5VJiBvLakTER_!tawLXFt0{DSJbtUnNhZ z^(Q?Wl=8LMj|g|h72vY+idHJ7k>j`5wm+k1EXHH-&ydc9Ws5QZ(39`1rM79xO8 zH=-}5QYpvqgHu77?KF{*ZU29feu4fb5#IR_969b%I0vI3+nH8BohU@ObkxGym_L`Q{U?L+ z4{$lYL1hv4vCrR#GL?%4hWTx$!sx6%e>v5}vGp&Jy}y7blEl~--?*HjU4JsspVKeb z12va7?7my~mMQM8`Y)TNj;;}S*F=Z%rrqgXziF~qJ}cb;x2?fR3%HZWrQ_7xtP&K_ ziuCw)p0DD!diYTVSN7 zx+8PU4k58;cxunJg?xy!uBTjC?@(xXBt?5E+C2YZS%%&Hbsnwt((xdFw%p3wmK6)d zcSl~G_~Fy4=YiT-slBKWW)LjA`730;ulV0)5tfMp@*o{d5P~&1wJ^+P35=HIo-^1> zt%Cb%sEx6akr54G(8-CJiSMBV{R1|~o)rQey$JQ8v3Zc(uRrcT>FIGbPedpl2hDs&dfce&6}aYcgQgeW z;bh~=F#kUC5g_)GzQbwmerdvS+#v7i2d}zcF@_zd{LIfhcx7l-Sr~ELAQk>~h*=G( z^pT|w&IBn2V+Fjrn#WS?dAw}iTEgFmntcFOXBHn#AV^vsh-oY;l5kH7WPo+|*>~QT zyLt$hHMM6SyprFeFvi_K#3v2tG;%I+cbZ(%8of7J+-!AAP+-VWywJRK)TT-JFrpy^ zgm>|)HKcT5SWM9u;OR_b=*!W&^WpE~)5W}Z7XulYVkxQ|F^mPMju(ev8<6@6f5zA} z!+-zyfE`-T!-C`3IJGu(oy!dnB`v;VU%fMf3XXtB4q&hk-z&=>3yQY8c!T7qtnwCi zU7Ep2RB^fbqr+G?)(6CQ>Wa1w130N?J?V_g*b1`HB03>*;U0~lzK7%zQFPbs4sH_OnzdHAi{*8n+% zc*14(kKC5uUQe@L-!jo@Yzm5 zQKl@-q_KAEx4d#n{yMy==Id>Jvawxi2YWw_$3bmL77E+znkn0yovB*rW^Va-vAz2; zf+iu`>X)Eq$zTpic{=f7*i&2ruz&PK`OH2{DD*tFrKAe!wj)+8T(``v4!;RFU2Nok zpk`7iul||i=*rT<0a|0i_I{}0gErYV-y_J|BbJ|_F7nxn}TL{aO^ z-B}r$C%1c8yVX~!s;bb$ce>wocSi5~Wi=oY&@+G|H?X^oguPrDyK%W~gIv$`7F^Cm z9(Ncw%65RQYH@M#O862ESJnFC^^FacjEzVfPXVAuVn@Dw6mqhWjTdqIORGJmVrh2P z&u)C!bv)kv)hE}{@6&5odw)vIc=bveaR%aqne%x|iw)IZ<_bPx>(bwImEjo7x&V|X zvL;c&KQSeRQ9w{II&<8angx+(hk)mcOG!y3N5i%8O!i!you2}0a6c4rKj8vmLo|H- z%FX9L!bma;OGqNVV}v&=tzU=#X&Rz)zO|I+1|#e-9#7S&-Bm%(3HzeZ^*%k|x#NGh{_^+mm;Sy! z*HaYa3&_5Q$aA__x8A!IXMdVGBAWX;Kk0g#{g<6~+g>ZRjO^uQvGxr; zBZ;MjlKSMN;J4bDCH3L&*pC==Z%)$S#gBgJu)`kYlRR`6K78)O-aO2Hn9l0$;^X(q zN8k8x>+1KQrN)o!^wtY5MG)pa&MRf>P{ z(B7eo0C#ilzC7=+U)QTZdCvZOvseViBsHz)gxAwRBNsd0UsVq%>T*7zO^WU7Vmd5!j8}&= z?drUs3=llLuJHw|RZRI{5W5r4gp-<%gBG>MZgW~wlq9@1fIW7eAJ`(*FjCmfrC2u^YUR#|AEvqaX`onCsIBh zH`vJuT;ht49ET|L&vzFZ|L+m$*&R@*v`+x~IG$hkcG?}c;*;)L>53Z}6`B`26`EIm z#|gPi^7m5M2hS-WK`mOWd&~gGb1Kfy`sL7}ZaU{y+(=YZ&*9sV)1#%OL#7#jEc&+- z3UJWOJaS>%kjCEa286x?1WC&#IzyATqB2#b-BW`uU2(n{;fgrhzEy+zxR_RdSZIKC z+ktmadtSmG%ZM~5%iUe?v3%cY%h*zve$R>OrVM$XLUD`iGtNA;Oan zK41vWson>~&PKO?ORGDljad%w3yh{#5=z)dYBzdVVA)ybsPKSXJX3h*k)C}xnB_b} zTmH(8TRWY(l1p^%bXAUnlyy#P*Lr#>5i#)=z#(1rVw9qmgzPKZf7XZCgOhdfzr6+y zHh5;DuTOe>(JNw%gqK*?&H8Vgi1XZU~9Rf9 z-#gb|2ck%7SUK=0mJguP`4%7}_S|d3hg7nNB279=D-@Y7i2diOvW)4tfGWaAt>kww*g}Ly|!M+O5lVJ@^OT zf6oDl#RlR2&**FuK&o`B1v(B`OT{=%ea!2(tf&KFUKQK%#j35LZG?gcRj>%sSoY&> zMv%CP{^dCAX3!zWmtDq5eJtRr@y^HMxUJNYk57au1c4ZMZI(uXX#}db)%`Ly{o_=T zu9PO}=;$~Ak|frn-sJV^X>nA7i=_t5_I=0fAIEd$qVWv=3?eOFSKgh-XRXx(ZRvGtZm7 z1`+B3;25odgcsM)@*6&J&Pe~_;^MpWpTCW@i4g~atfS3sJN5AX1Nrw10-f&5G+0Rs z_K(v;S&>~S74>Tsx#$fWBp)Q#Bl8ayz2j_E#($!!P5cm2qjaEGFG5>|)?{55%=G*> zdh|qx*5bB0h3fsuxq`GT%Tg)_JqfOq!Z*X8!;oFMKML-v5q>@Z+%s+H^qXOco-K@^ zw3Xe7t>jF$h`5dBnW{B~6S*ZvtWf)FZ780by;Mg zol=WJeQcpF2PxR>LezY30XCGrA6B2wYf#m+ItY|D=0jFeA`|B-ZG9H+BdYS}c8C3+ zB4~P#=!;nu=x7UWR7;c7bgW}Ni4+ZqmQ|Jg03ajYe$kzY&`Aqb`)yavNY4EA?eSn@ zXNm1Xv^Tjf#S0ALt86;(7e7h}K-+$C$$no|hL`DtgUqt%3uWo@-w8@Y5#OcUFDHVX zmgp>fnNjc5#_)ka)F?dGqM(%gEFHB|8Ikn4!a57Op4UkEYDnT=_gb;QcuI*x;@V?6%rWz%ug z%(eN}q$8s8ZT98S>k^H&PwjJ2FrpfZeQ5Oxw>5R;_ab;qkhcEH}Q-#({G51DnR)~fvZRhHc6A$F3Aw>eH%E? ztBd7oE8d)p+DeF-8DUb8Tk5c&wjZ54{m3lVj~XCkPo2c%Chs4zis6 zf!&U;)&<6F#j_c%#?{oG<`KtRucx*O&;uFHytNj=22u35Xpp*1Gj9)jPu`2sulMGp zr?+HN-m}v(zrdijg6*h3CHph!I3rRfHBzxP50I-7fqB@F!;fj}NzwW?dv!PD^4d@C z)-bIIk;0cmoul-D)}ipoc)MRpQHc1I+Cho}_RMO>t)4pgT#m-(%TIzGx+PMD&t|IF zX*YFuuG)ek2~v8o^Dj>1C}1#3tAc?r5VSx%$o+v(xS@-4y* zZB{*f4~d&>!x6N8;ukjMDG`bKFyz|GDBP0zC*zrq*25rs?rx8GMN+^0K^_C#Hclde z8R#m(;99GwN7;uvE|vU0A{sJ&mZ#R*!{n4&dCD^*Y0XWT3Mtg2SV1$+5OrRqTi0d5=0W*a z7Xewzy7%t9@&!_GF5zga>K(sK;e)`$NAeHrc^+>EF8F+%?$#2$;kNzh=ave=A+2g` z?)l~et}{r=pA(QK*Rgx#3W`;NQJei-@LA7bbaTGfdZ}uwzL{^B2Af6Q z%?&ckDNA)-Y|Jk1V}&PhA4nu)9#2=qyUi9ZSC;Zt%)Rv?*N)XewphWujq@&24K75% z+G$%~Car~ljr2!-LWl6FuHg!hc2ZlGUWJ#vy3=o7@atKqT!hL-+*V~;%uN(}3q}V@ zk_PkHzNXo}*_)9PWo@8!`1Pb|Z15md5P0KVya>Y$)LKK*^3ayVu-ld-;-+6?ctp|9 zq1!eoYiX>9je*9$u!+&uyW;_?UB7|_WY%UtS5w090ifojCh@VMdG-Q2d5Z?|$^ zi=O;8%>~lAnV+wovgkt?ptUx1Fprd>Y(@c`{tKhC6spic9+;@RC8j7>)1M5CB0!stF68;hWN|u2i+gC!8lI)wZBeWI| z%#Od&_(`?q>T;L#+x-mZEfRI)x3$E%exsbQJy}7vmS}1j1r#P}*RQGJ*)S*qJ4`wC zz!$ODyVWfQ2%s|?(%Q@?u{0S;)LXr{)*4tXMMrsC{tw0f{!IN5po-tISp5O~A3jxT zWhFUF?rA?BJUM!v5p42LjeTH$SJq_AmCN^Q{m*6&8GyK5M$n~y+32jm&$x8P1FM7e zwND}!FN>iYyc#usA!Y6q^PSp@M!n~nhq#G z+IibP4jkbk)4SoP4iasw;jw)DsAaa0;W(7P zWA;C9b>O5xkVtB|TZEy9gX_FrhX191u_zLx_cy=2h{P^euLiX*XigmeM)$UH^35D zrQG4kXRIPm9bs+ACV*Oj=o@EjDCP}mqe9$(BC4J8$mTw%<4)tms(ZL#!+`&G@AS`M z1e;9F) zS2eX@R{1Pg)zJXc!RBth0fcVgn`wI7tT#Zxv)xnFLhOH_3XpLN#KmaEv{jCsH4n^s zan5S#IG_h40H9NhWgI2>gFSn`1-j$^5|xm-+QlI(LfSsyxe*Ip&>@c@_vUYvB=Zje z^#z5r$QI*%hl=|7&$@EHkkMYF#I2zIq`OA);#yi0d_#9NONMXR#KvBPUlIH*l5jcuReLGz8x8q5n$2OZO+@MlQ8x-TuE2T`9qL z@-G>D$(F~HV;f;iG4auDICFu5%`;NO)|D*Y-MMV`tp8<+9)#Z8oUR_ZMYUt)y5F*XIu9{9?eCbhSwbU)!+K0z1n>En3ryX1=%j<0x|xMr(I z_LzlKkT;Tl%Y={QGYkHAQDeS~Ln@Ap%^9-X;&WAdttuo6EjspWGDabOR(q^Zzl^Cg zXn-;#Pg4;0%6vpB1|C`Qk(QMSexG1IWQ$d4v+ADstav7yMq4ScY$5H9VR*@#MH!if za8RR)T?#UF(vp?-23j_pTX4zg+Sm|0TeoR1@>)(=W_dRUBB*peX!fZD(5Ct-fY3AH zc|m|fR0!8OSWpAH%*vFgtrpP#KVV&;{oq;OHGljC%6`&alJkm+v%9xLid#4FPxFAHCS!x3 zOkEQCn=N={8r0pD4G|mCk+aIDJL%s=-^qS6<2%rq&Rci(x2&60($_SySjgb5F7A`M ztYg-18c3%xXD;^&k_2SQo!Q1&a!~%w+V-AQu1s5Q*W7KP82jyIP~QiemDBf)Z2h?*y{Yb z&EiI@LhHiCVwVXut*V^nxdn-Rh5>wdsZ!e%n*T zg@!-z3BB)^S)vUjwM#Ckp)o0^K%53Cs$Rmt>u^33gdNOYa7I#B(59Dho9zFPTi|X4uqBq&i$F9;N~!_BB$3~7xTE7Dq8_$WFY31akWTgS z!|zMMF$;25lk(`0xdIKB|cg2w#8t2&z=VDt}SO{qqq)+XQBSi?)qlOj6u{HC^ zQEo<&$0imQ957;@$X7$DD+;2ow)c8mTfNc4xCbTVkA>Z3t*WQrUopbEf`8R#uU_K^ zk**K74C?WV63ZN;-8U4@D59IUgk~_KOm4WFG{)7_5F}pAdwm4&OCFKEJMsF(E(0D@ z!-XatKOVj_-+C}-cIm;HcVT<$NWDz$&GC+aR%tvtc=AJRfy=z3G_fhEdg+TNJ3cQk z7_+Blz9Z{sy^r`I?l_d-QUj7kW5;s%!UKVOP|uUHPR1i&x_FPVqMAW2vv0l?2Xzc# zwbcM>Jug&RF8N=>71|4^BN{~s#T-L)RO-M%@FY;{IcnUVd1~uob->UyN5Ijc#0TM|y z-vR$qh(m8$S+b*$h(|By+b6FP5(`rdrf5lDcet|U!GOuIkJQo&CVqUaWDW+SuJYJM z*)+3>pzPL@`-{{Xl9Mp}nvMfbWgA5z%t9N_*mwHWlC}qIn=CX-mxg9o2c2_R7N&oB z1n!w;a+M67IEq=tz*k};f14I?_WBgeN;oJkM)7fo?3%Oh{D@2fB`26Vdnec&!E{#G zo_0<`I&4^o0eIcRHvVQ#irS^iOIO&gi|+P3dO0zw?(EERuWxYBBX9R8wxO)R>d*Lq zkA=IjXIv8}YWM1r*z3#FErE+N?zhA{YjYOaXcvcE26@H!5*2-O=v+N!O7C4OcVAg! zu3r91Z>FaaW7xT=?X<&ocHj3sgm@L8%vH}YVs9pjomt;oEKj%}k-bzl>iCGAk$A4C z6CU|>JXmV@8*$G-_-ff>wY!z>D&gQgq)K)%*5Wy|)pJ;9iQeBE{Pali}Qc4&iw#f5qBDk;q#uO z*QeeTG#i6^umYy${VzyEzxCd5W`tx?irhN1`~~`DpiySnE-zvMwT%Z-S#}PpFgFpp zGG0Sf?DAzr-8m+~J4I&x3gOd6Ig}wyDzNzX);YE|8!Ok23Sh`X2wrCgmLrqs-?1^S z0y0UTjnXW3x0tvgmm%O@c|yyvF9oR+K?>v0vQ%-~YYM?C(&w8IDpoz^9WDQuDu1{W z8+!124a1%&0kZpL$=BG7x;z3t-?|AjzknqkgQi}I)I|3%G~;q13`HAaW|hn8PzT)< z8`c(SftmeT@gv{;+8)2k%tA~r1(6u}iuS z+RJcLn86g={Op6b#`gj!Zy`PfORl&r1%FKnmb6B^U#*K?Ex1EtJyrmml#)IHO_}fA zUt{xSHED$S=6_Ie@SXyZ2gO@AyxzcICJSt3J+2xBxmOK*nCMG-Q?Gi zbHA$=Mz@GIKY9Z+kP8%l)YBOVA4*{U5%X<@fW+kl*9DFV13*`R;HUNo*f?6ub>NtE zVD0aPwz<1UxNl!ZdLE9R9hI86z#Rozs8B3-{3u-a_d|s(6I*FeCqBvs`wY&L4U|a- zQB}}!I>8kX#s41g4^IDBZs79K%1KRXn7{J_{V{xG+)E*lUZ!q`lkR9xOS(_}d}}6q zYEzs2auSG|J#s~t`ECYbE$#`UpA;Jr;MfK9VV_%G}ML0&irH5^Flf_+PL>c(qx3oh4!@yaG54CUc!2)QthrynQ^bX&&H@U?H9n~h8*w}XTREcE2C=EMFxojQnfdYVDIT@!%VcpV2;!(<% zxZb~#f7`&OR#1;_C4!bgnl6Ft_iOR5#5jAWNB{H!7}^?bo}h@VyWpE+iBND) zMR71UpGGZCl3;e4!g~T$TsjY~F6Cn{$%Kc%TVzkKqjz#<(7oQ7mS!wpir0Tyi6>xW z8&s^?+ss^YV0WBz%Aa<*X;+qpDWf%$#d>_SuxWib;%6aW-AvAi13E?EDn0YTdVC z%P-7)S{Z*RWv0WU#c&`pb+CmtbvKR08BSxZ)WkQn8sNeibwWSGteE?F-#MVaZfBv2 z%$55cb^^`(=~R=<#eehkr+fMx_NJ#fy{nNbMRooMPFx^65T%VtnO5Vh@8)KFy9c*E zIfDwXmj+#Vfm9##3=9h^9ilTTdWw67R2nvk%?FcsI8!J~y21u#GkZEyO=rnKKx!XH z6zMhY{VICj4Cf@x3muMuA7LS#v-*VLPYRoOY}Pm1ibD6cRpSrt%Wc-%yA0Cb#}lOI zI4dOP9Quhb-5EjdMVU$LV)&oc*t2oYZ`TlfySmJl@ly6+HQBWeu5`gaeeP;o{+Fx|u0KrLZ8yT( zcN?@4m*=@d*cf|&?hfb{-U5L+OnKv^jaVmKJ5x)V`)OK?gY&w!wZ{68wF@@0Wkb6V z(o)x3yd~0Jjv`9JgImv3B-*FxH0>g)rxzPJG)TQoI+Buj3JMo$O2pl>zg+A|g5!k$ zQJIqTZJJSoJ1Zp;^Huh>laAs8TgtWYMPCq1onN2z^>PUOaLWl_@H5Lk4viIo5kDc`crWI*;XN?oW!EAm^jWA zer@@FV6I}(bCLe@LOls0_4gjS&fGm!BEa83^_BczLCMKWfDCkRJixyL({l98uJbW6T_S#aJLK?d1~>HhvTcjZiRlia;*( zkCr(+?I%#IU-*ASBb3+6Lic$nwlTqeWgL2UDVR|vqPAPhidJy`vsmRypq2Kd!-?$5 zsVK9toQg#%V-=^~D6dm&{?n6Gthml=&3p~hB3otZ<#r|-_Mk}2 z8q_z`Ns#N1S-H(=jd#SO?LfN?We)f6jFGjVO5%?3UTU-@E3Uw&pi75O@Oz0Hy4c!6+mnhvahQk`2ZLTf*_o2}e0vRp-SVOF_RGSf7Y(_R1c zODUGw0HO6~GSKC(XWQ*ZgAIOEc?8bP27M6XFLxMV=_7J8n&|JB%I2=}C>JjGSes)O z+BHva0bvDgw!R@dNZ~?t!>5HLaBX-v*28b80!{Y2xOS)c4%ZR`pZ)Bog|rY>*3C`H zB@2&$AyEv!Kz#JYDj>{u;30U~6$@9&f*uBTs<<7!ClYy*>IN<-A&RW`=c=Zd%Z0hS z?v2w0yP}uU_<%7o*wmyTfJiRg>@&-hf0g=r_IIO}$!8fx=Wy|Q$!E%E(%RQxkzuPL zDX*WU2y$`w#WlH83#;i~p`}Aj8?!Nfm3&N?LOCO&fWA+5#lyAClyGfYs^`R()K;rI zRjl6j;|Zw`PvYqrIPt-o+7&KmA=oXF%~3=P-XX9MymeKUHoQ>HV9sm4UWc=QdD3WC^D)i5Lxzo z{=HQ?9-Jk2bq@)p+F0}N@arq>A`>&i zXy(cmx^dOiDqegy60Lf?T7W68gJ*rKix=S}O!?^hjl;XFwPB1a{CyVN4Nvj)=$K7Mo!O}rMw!IoiKjt?To;*MSsh3o|LD8K@ z8yk*}5=%ur%5=kA`CG_YBkj#WYZn+tCQR$))W?}jhlr;Nj_bOP78~Jw#?OYUw$6!Y zIe3eteS}u=4HtfWeAG*$Z^AooOVoCWU=&*~HwK#TSeSfOaNPN^#1!-5e&tTHmF_L` zwgzIMWr))tDD_kuatBpeA@37R?jX9<^k_jhAF~+HMj{W?5TDm<;Gd4UmWyqfF5 z7D}%2WUhk(-zzukzpjvsh>EmaEQ@4G!B^*iZL(Ldy!lWusanKaX1Cay&ecluQ>EZJ z;@z&170Q4NgfgOLp5L-Q*Ir5oNu?7!yuT8$d5Hzmv{MImh`5Wt9(32y?o~~ zulZG;COl9qr5zESyTEYaLHyihE?pte`Y6iY(+>5@PAN*M#_`n@&6aeoQk*_d+zGDm zp!r!gw;VMPZ1V8bw87gLj&|p$1odYIpjO}H3< zL5yx&&@G;Co%K`|o_dTVIE_;PY(Z>_aq zv)@WDCyg|+g&~S49rOzM#z1?+qlu%W_KM5=BaNzoVbI8>R+8GHa3KGOmmMd3-CwC( z(@jQ3=6N1$ocUeL+6U6ky#^AuL=aaa27wAuXDT+F8w8vcoeVcoFJw82T$fslaBxs}sYI6f!s64R(ws!W1VLg?|d{)w(Ca|Gc$i;kL_=h zl(E=~?HHHEvJj$RJ(Y)CPYVp2S%**gX@MqFKOT05Da!44tP!4gCw2{VtN|TL)GHBk zR(@ToS{&GdMbG=})CQgs{+GjvUBwSfr7ewpNb&h_15{5B)QYj4DuYE?35@ZHKo$KEz=zOdIl{YYLaDGVF?$(UGqRDN@%j7Fod-)u4el*PjlT)i0Z5Df}so&oI*3DOG zy&RsY9;#;$2rgL0W%|yPux>p6u!~TWnOZ6~2UjnZo#cD85d=Biv5wzej`7_%Z`pkhFV$!(z4(?n@`3Rj!jPf>R|;D+8~ zEBp(Td0cE%h$5q<+e=HBod*p-tkD;q@-BEVt*@mJ9Yz;RHdh&+i3p z2@Gv{HdeV_ExO0Y!RX3F>n1yKL*WWf@<&xW8hg8QPRLh5K_BdXGF_W4yxzRt*>=#+ z;UIVt1;N@G2<@0i9S03<(D3X8Q9o@;CNgz3s;APhW&=fDtyq5ZVfjcqAGj&>a=l$@ zH#Xt-)u zzW(Z4CW$6_LE`d1Bgp1c!p7W$EY}c+sQF4TjMcm!r*Bw`!%zmmdl6P4Dfj zjmN<2Lgnu!ggtoaG&KWHks04_55#w4!3fCkrmksFcelaTR=j&9C3!Tb-L~#|-b6$+ zgPH(FDU9hh;u|+T;b4yb+gywPW*V-_V{V((x^}3 zbgN_Cm{DBNL++k{p!aI-{=J;9xnF&h_*+$i2V;ECH`sgCQ=kJA8Mb>>8-;hpi!*WTmyG7=oExj#Q8zLuV9@VPU3xa;EHGn*aR7-DGQiESstPxkkTxh zclLa1Zjpm5a>PffYUmQ%&s^BO!bnT$4Lm^obs#LKhPcM8Bt5Nc!9t77IqGhV6Y|L7 z@#t&_$DUZ15=B0W*>9ruR-r^0e$rnAy%O9&J-l7NI&O-7c%~Dp`9nyPw?IvBK4mh3 z)r#QCOMQNRf!?x>jD)}a{zDo<|0_PQK%?+MPrh+qh9KW>#G;i#GTgCEwxC;7z{eC@ z*x%*60WI=gi__AFJr~1j%@i8`B(wt;p6;)w$pg$!*R0DJrGE?9}nc? z9A<5jTy+F>?ID&&ZiD=BZF)(K8$pYsHxRS`_%j!|ZaOXwl!DUyBt_<0(43LnEbz+) z>5b}0&9(r%ph?U6M?FUWb}b2>ZRN4$tR<5#V)x*5wAL+h!0;kjw~ShG%RXfMeDJ*5|YCfYB@MK82C%VjrGh2+*y-<0;|8Jn(=*4yDu|XPsVF=57;L` zb`_vE1vaVb(uo9H{^GKS$Cazl&P$h-4J)uvb7Y z45K-6O2_<4r{i44z)c?J)^ox@QZh6X%1yqZw)7Pv;=evOsGE+mu&}sg0#9DGND05p z6Yz=>dY(H>;7=IqZ`4v~>VvfU$|RRZ84cEdXaGgF7$AG~SXs1M_RdcSb{)%4Je=4cjnO=aFgp%bWk67c8 zooV1c=_au*wwu^APB#1Q7h-qCvN_^Heb}EJ6WqGeFqI%KL!KY zEZ7wd=%~Wn#7*&CRx=Tfh3<8$LtE8+Shx8B_j5INlZXW$k@bXa zKrfUhbj(d)s7-&+fRO3VFMY2|d>b#rK>q{LAJB~dP(o)xjiy=lC9tGY*;~-z!fLz~4mtK18s z8%scJF}I>+S4^g6D~2|59_*$U1A{2J5(DlXRLAUT0~>D)NPwU%7zQ;-8IvA|?W|c0 z?lYDyRqsY*Iw0UFlbh%z4BCo+=i~d@bYaB($0iZxz%a=$xq)B^7JBSHAD;)}Q?!Cw zy#I;M%s!@PJlB!+L)m>JQ!o5tkjagwQF+@*bSUHAC!<9?==U<5Z^N{!00%Y|Tu}ga z{`N+$2QAjVNX9pnCQR*3KwOhWHJ#0)wO&(y+FDF%Z#V$Le!D?3(dMsvH3T9T z&XRM^!*-xs*gJvsz~j=<#pDIMWx1YqnKcj~HWd&O~1K=%1F6|o4bn4>C>Uq*oWP+Shv# z@b4sBw#RxBT_983%fuLn5OO9dvTkn|hH=?Rl-+R^aR=L!5|}o|+y;j3YS(l}nRth6 zlL83!Oj2gDh=hkKh&c6*B3M|;#-41NY)+<$q7d7F|McYZQ?EaZ(kah)-*R5_;o}=# zVYS+MmAivWoOHAChOc$4>n?@7kK6a1pTNM+KRkNW4FYX!N-)JHlIGICQt_=oW7F%| zw~O%%-s8lIdbn4v#V!SaK8H3nwlwFuQ?UMa^|CkDx#GV6wkG_4_`2$_D7WoRj7ms@ zNW;)WiJ*jZcO%^b5+dELNOwz0H-dCaHzFY2Af3|r+njsOx#yhUz4v*(hiB%Gnb~~b z+H0@%zVBMgM-f|MtwRPMRSNm*UFHYyv^oiY4>c*N=Q@+KdB5(ga;nd#WoB?CKS z7>V<;gqNGoI#@cD<$3YU98I#bg6TX?>*tyHWu{9Ku8P{Tvkau%N$vVsZ7z4JyYy}j z;|~{?H?%X&;>=(m${&*I8R|WuUreCcd{eiPQ`Uz5Lgf+0#a(2e^IZ8jD_V|_eRsbN zBmA~&&ZUJz)zRxJo;_u|F?TyuCNeb78%mCXPVA*k7lbJ4q0>^BQHomat7Yw-o{>&R z*FEPpRL+6BE_diP_>Lp%&BvK6$)u|0Hrkb+b9QM5vL6rX*SN&vy2-d~A+})@-Y*MO zB-1DbeL!*HSl`|(QRKM0{`iXxP*vo*DnPz9Lc)IlXC<<#G{ZeXbk40R~_YeGX*g^uf3!b zr8_bK>mB!}K2(HqP@{bJIImuL7@I`E*529Xk5NApdZP2f_On5Vw1@$vT0!Yg$K9EH zHKE`d7l0yecE9W)f!XO)ypC+*1))WrdKt8&>B5`@57yv^$W#IGl~YeuU`5tQ<@e$n z+y2JqWtjqz@cs{%kdVNkT5tKUo9oLhl&!7t(sAh}YC;t#0b7!aW{i>s64!JIirLW3_$`FU*bR+nG2d|e#5FN*Avm|i8Wu5 zj>UzyhPwPdV64D6e(MDG_CgN&5?Z-qmCjKDl;E?~#>gNfl6I~yf$>f{O6byZi*vQdI z-BC9X(zQ>oRb}~s*;wP|bl$5p)6DB~hXd3EQxc$f2Zzzz%m4lzrPFETVKheO1Hn+h z_vlets;RgWBICY92#cnT;9jQ9Ds-pkOiE`}?^P3c)g>3uP>0Vr%zA!5ba{p=%l(XW zJL!3WUK74;bN2!h9s`Bz`t0Yi!nC(Pp@_1HSI@rxy!Oj5CwM2y`?7h%rbOLFL}#2# zk@w4x8s*x@XIPZVrq{klqNs~JHbQA8^cvVzj`0+SA1w@+bXzTzb+5PG9Jj_3J!Ul* z^de_93!g*l3#oB8D^1R;p&8?1tt;ugo-9gyH<)49o3Ii(CPVY9AKVUDY}@ea(($#Q z!*8Pn(2*?VCO%YGd4%wz4+&>13)$Xwc5E%5e0*nZcMSho+dew0p9Gu}iMM?={*S9pw*fvUWQ{rM`o7(AfR~{t6 zp`l}nU$D6tmH^r$#tq*S3}Yy$x8GPQBp}6z5Ky2L5QTd)tLZ`cvI-Ok!}+(5NRWHs zNX*y32g-*FUCeMXA1Irw@VG+qM{=MCwWbT4j_i6ePd&QOsluG$PL8*p2Bf+2`D0L= zcd$w%@-gZ4cYJ~j{R-!KUomf*uaLhIV)PQ~usL4PIxlU+wu`Y4slzlc;eJA&HE=db z?J3137aOB>5jXz9V=W9$bI@68ldPF?!_U+}?~{mQHIh-+u8zd{iV;=w0l&nNndj$G zE-&(p$=t<>+@^;FOSuoVMc25i-iR(4VNiO^e!O~u7seWbEHib9=sH*r&rfnSK<(LU z%|T!^4Yb3m##KAn^A$4SWte*#%FG6}kX zosO!tpd80hFXbj^r%n_gaLk<{J3Cx-TwjpiQvTuPyYYcH+@2`QBrRkdk3(5Ck%_Y= zqEU=Y?a%An-AD+?>?~NEYjmsEHvdtyfsMcFuA!5hBAfX1o)VoTtXR5?rYs6J(H=mP z^2{P`@ax8gyV)to)LY(R@CATw4-?VSA3V?q*}sKTd#&{+vau1{d6Dl(9lKz$h=)A> zgb^s<4jK<+yqT`5*{PbInbnzwwl*F1GL&=y(XhiCJIXVNFU-4lAJp(?igUf0VozCl zb|!2ROoiD5A%>!asTrK<#__y}mH)mam28Vs+-f5ReM}R6$ude-KE9ZJeTMiwu265$ zz^cbc&s$maC!q<1iQu$+zRUqNir;RiLYB5fax5IVfws3U5-Q}RPQDfAMNF~LtkW^UWCY!o< zy-ls<6Zoa^a-Z%GKPcdgeb*pa2Qhkc585tiUc4=XLg4`hC|rClI*Uc%OSS9>!aLib zb;url7fQe;)?uaHJx%FBNYfGXaM60;<1Br(_j>NvD{R#Z;6b;zImJ{8 z1Nz-gw!hj|!c-=D8h8?4*yU#nqmuZfqozj@tKDEy*;nf|dvaHB+P9hZQDN#rl$tKZ z%DFCyV}5-T%(^C+YbI2t@yrr(>t7|8JlFOk@+-?O*0wP)s|;k#VbMkE5hMu6zTu{P z*ZuJg(dN4s4pSJDq$iFo`6X02GD;L0G>jgTN)y$V&leM&nCZz}Zfv8CMWDVv1(FB9 zlRDuNA{=HPCLS-VtQel?t)o4*3%hy8AAY8A{cw$^?pE#|cLLUPV~qwz9gKFRQLWoG zyMYugEsywlkM?Ai#y?$ItG_p6TO0zfA$IW@=#G-%ae4BX%a#h%_!s%L&wJhKu#D|a zSF?P0NRSbX*3&x=NRz%w_t}vUtEO`k(49lFnxQLO;;y?wa3IbY_9J7&2k4_%(HCEv zZzW@0C}T4MKHZ)Vd1?D*GLFS}%LNk7mKz82Ed#2^->pQJpgIy&rF{$(k24A@q=s~1 zN>bnEi$q*JrC(=vJ$k)U<+P_uGEi$d*|CtOQ}4J4k_f*LQ5mRWDBsEfiPH>)lcstT zNToiz*mcgEV1K=6LBx1X7O0;MSD^TQn_nxP=@uYm2`|QfRWEB+ns1m;+iOI1N~yoS zzBu-HSrK)Giv3)M(&tK|CI2vBV!Y@%?-4kqFvl(lM{a6&@_vD4>MF>tCy19=irEE| zY_?-8Z+?_|9@7e$^cSbb(>q&8*m{EFjW)&Uij493?c9eQnA8|=)R>x=nYj!W8sUX) zU-Yt)N}tPB!bTIxbamoMHYM_7$jh`j@07LE#~2=S^0WMiYAWm7BP=Y$S#ZPrN&Tlh zZ{J-k&CcHPq!c>o&^f=V$~`CY(ZuU_9SQ#h+EAC*?WJBdZz_j;z67l2 zfjJFr9=?8xJu1~_As<=*h95%T@=BU#`&uE|;vuUyRs`s8~-&64KZQcy$ zQn8`%l7(2Wp(PCv_>(tq; z$*GMRfy}>>oIWyTGB7+U-oEVv8aIk+5Syf`1C~5mw0=FHfi=Zny27TY)}$R=16yN^ z_m#H0^Io#tlxOMKGg9j14VP5A4d=g)3}`BUsGGqlx)zm`P~irMS+nOL3F-SDlgyBj zpY=F``T+W~S5~IRadT7>wH3s^gSO00ka&R>gxA*oUw=rX!(mz%d?WVim-fW-`iM-V zSB-LDz(!+4ep7xju6Lohlsq3tw03q`cjSiclSbZme*K(`%k0`JoEFy^0iDH#pI3HH z&rANzd-JrmQ`ilo&VwVzpQijiAq;&2weS8k*dZis&-2v?t&dVZG``DaMqTzL{ZP@M zb@LtUS~}Hl1DfL{I%UFmqj;a>ZS`IjsVkN~jH8p)qv7HZWsgY`v65oXN6G;{{0Bpb zhp1%GH&|pkwhhc^qUueOt|m)i6t&bybmPs%RzwmoqU^qwPbH)upA|}N?v!d! z!OGNGC9REiRUD{?m2!%>^f*{7I;5<^-QQ|5eAtU^52({9jc`qgytphcD9mEV-uG{z z7G=p=WpxW{zA>3b`=Dob{0OIb?jyB$E&&5J8t&;8@BG$Ath5_%@@jhXjRxTY`X8C9 z&u4OSLme>H#rOH6o(;2=$u_Qx~yRkwSoTHM(DjH4KNb^HG-n{_d&I`O~8HFcKKzPY+M=MbDYO&!tx5yF9tovsDqn`)ri z8Ur*z1}Vfp7^{I@ZJMir_95ky%Io3?29o5PonJS{bN2Wdf`bf$>*}1;2fsd6Ihs99 z{1|*Dj;8br-6(kbJKm#8+}03Lc5?);OvX^h%!?DmrJ%1gpRz4%K3L!0|Ih} z5g)i5)HqADD`^Ewv+N?Z6|cY8W#=tecl+?WUxzYcIhPC@bqbfB#>ZI(ZNc<*#%N11 z=0-PsIX5hbZD1F63gUXYsp0s6tX?HZ<)H|u{rfXqr&0e(1^@L|014^rrTNwN5%2-6 zX9vq=Zq?^dsMqZ^2apDA$G_fO`@AlPP%A{9@$!wxeCB)MSR^hjWiDAmQUc`Q-dyZK znqimWHnO_UMe}_EZ%AHH7KqEUkS^NLot&2`5q{g!r(zO3TFz{^aDMSxbqvmk%yZO@ zUka2A+>jS>N!XH17QcYH9UnZ$RWBSD=)cBrJx2C_7g;<-UVi$F&mY@iO+f@wbJbzG zY)`ijxv#SGu4~f;|JxQ`aj5)sm>O2J;H^2XN>l7;WisuB8@`~Cs`YvFIHGazgEr4a?;Pco4o8D`_1P=d%dI;&;eK^f{tFZAZ zPw&`VX7{}ate7xZpq9l$;2TcU1DL$uC)U0|o1M-!54_mg+%f;oqhAZtOtWf!5+HG0 zT+(=d-`Q|@QH2X-V(tK^pD*_ z*A1a+&c;2)V$f@H7fgk%#f2lUNWkrk6uKC*X@lt_2*MMl!Mq_?GKe! zvr`i*yvQpS2w2a|p(kau3o=r^%V(_WSqq`qwkv$i8X~a;bcf#yt*L0+$~=-%l|@px zBU**d2Z~@$Hrtuuh1{xg#5(1Ve~je9;{|Gz)P_Hy<)<}gQbT7wox@m5$+s?8o6nb` z(&~%#6AfJW%08W4OwN%D+h5^yz>b^XJmzM|HMgN(g{En|D*O?VB+N^Dk}_?e=v!84 z<)I#o+hfYMXx)}5P<=>3QLQ8HNO>~@->zXaoMEmklHa|3WuiK&wU2|yj4Q&OG1t2? z)YKrA&mbD*Yeu3vWl)>Khmp+vK*AvVkow&u;`E=ls(TNTWqbJnjp*=_o%04kj$GWl zAE@KPOa?yY3k_@ZMY>L{g=VV|0$?hQ1D}-A$6dT7BTic!H=>mZN~)Rkyw7P!&=Jvb zwCG;#l_BaUI;L}OCKw^-sEMK;Cv>!!JV9hrQ+0(k6e1*<{^}KSd#6qfIo7W9oQDnO z*DtPIdb(Wvx=K315k#SXoSwgJl0!-#V&rs?R<=AKbN62Di?9F5XYqrvy{juztyDKp zx3SKmIXASsTLM>cn77h=tTmKlE z^Ic6ImkOVR?#ZO`5TFq~KbjzQ`rBdf_jN=Fkl(F7iQ_GLt#Q?ISGZAs0;}ukM41sQ z2w+6S#7wL$#tLFRuhuiemkzqfxKD2=We@7<3G(vP5{u(;`(=dF9agYk&V3}sK zb(N%}AWrlj$(6q(PG?L^l=tHP9kTt$L;2TRZw&-Nd;u_W4r})GWM_If zTN)p5@g#D&Y<r^ zpWyC4@7G^9x(_iXQXrgpr6vWJfIm7(5NNQv-n*|vqH9$?(`eWc1m;)XD`~!QJ=z+_ zh>VQnNrLUyL3w4m2qRT8ri3r(>dp8P z(htnz9=KK`1a;tbZ`dmR+x8)@9QLvCXQq|N#?O1RoJ81Ok%fax+135mN04-1W#4W(C%ltqSFi@evd^mpNM+X zQjO~EcTU9EsN2?(A-%>xF5+Ys{Mn#7JM&bT>e?!I2eWfnfR_-jqG1oBus0_I_TndRN8;6EO1FzwmdbR7ku&q%#zYh~frod_@%DU; ztl{dg@1tTMHnldwIzaQn6WPW%O5hl|xj!PjhMRFlL5}ctLO4-nSe!6rmZJCQfpT8x zVOYTRaVy7uM#hsh?^N8-Gx$#YqN>g1ra|&;-JN3;*u^??Y4fCZ*YpTV>kN3g&F-|=wZG=E06{P4 zmiuC&b}p#4!@ec^=dXbOdjf(G2IsMvpSutIzBxB`t=R;SXhx=%NGaae-`#;&bIViQ zZMbkqGVS$QN1kT4c0;#OW}+kaQ2y$%nOWkAtJk_hY%FCE?!j(*=G{(-=!1l%+7uQU zW$d-LC*y~+4aA*273%C^R_R#IWe4QvLSRMab$~qwsMo&v?ZwNN!7iXpG*Yu&CV3 zu?y9diu293Cm1oxpuF2JE027to&_7__L7W=wcX ztLlB9w54(d^-CJN88)l{O2MR6)ea;B$oVK~QLQX3_tq=F=0^zlQ0Q-2*WQpnXQPnr zkP3J`(H&y(D6M;!w|u+L%KuE0k${20bgl3Dx5tzOxzAqkz^w?zvD=nhAQ9CT72?KR zxYHKx|4uaB)b1CPDg!F+su1F@mi2$Fdx68}$XGh^Y;;*7?Ztz^Xj*36_p)*yJVx$Q zHJ?6@>THHLH7}UoIfP&%-Mb5DjGxG;O%{f$ zk&)hg#HRtOuPi;jTgQ!;c9noq$YxPTNe~~9_UO3uJex?N)N-qV(EQVulLUwUoY=zq zaDa`$+86VE{LnFIWqx}I5y6mct!nb~cTTFlo7;B&(`v;cz3t(ccDL_pIiK$7kN^=t zUpz~f58?xbMpp-R^HHQ{=A-+rQ6i2(yVSM37ULzMkC-&mWfDUQ&p|hW_W)zwa1>#f zu2J;RXTZ_v-;8@iE;|;gUTsAJh8{zb20nl+;FIU|g{xLt#0I~a1%UDHASgcQa_sY9 z=X`3L{RKhG$?|*UwpyL^=MqHhuNBrQl@T8a)I073!6V++zqvZ+^t`I^gkt01fR3b$ z9?>MKg)jXAKH1)p^W_Z!N0L{+!MwIU0$JF!pKQysWG+w&l(c@3jy;g*|GXHGr^dg< z8tujiO%HR%?IH7kyWET=9SVB=4U-jv2cq#*{z9% z&F|l8Iy6I?%4Oq>X0S97-p=3SIU%q=75ita^V|nyt6EDVLn8l*Yn&} zy<)$R(5r)N=#1H(iG_yhxH$6Lu1J9Kz%L{QaARTc?r9z;1TfWt7z4|w5zsVEw-vOd zd5U`(rqFzO1Dbd*02P@7a=cMjSO>^l(ueRPzkoSLu3EM};NGje4c0vYT9Ep7nM^h; z+UBz*R1iA~cMmLBN3I3R3_JFL3}P8f+_s(%y}mqM1{#*uz2@6ggJzzZ zl_=?bAcf4ZTN~5}!;2$er#)CXJFokKqcWfy*SfhmpzRGZ zs6^d*M{65zhG6_2!nGj6U#8AlZ_lqS2CV69px*S@!5;x~!+%Bu$sG`94o7R7V}%DG zur&Mn*=cVsqHEU?$X@S8Ox2wXfXSyhlq}RJD(&J}qSU>1fQ7z@0S$x&AHWAafg$7{ zQ9mlS%?P>%DarasZn&-_@9S^m+I9oaV9q~$$h@wwKsYQ8g1dbX(QhmW9^fykbW#l` zgPz)@y2eaqw4yV~JT6bc%xJk7m?{b=ph>9_&QPKLcvma1UV!k38dLG-#kii=B8W}D z4yL@Yxjr7Jt@*c9@$URXQ*%O3WOpTveds+QyYkJE(b{v4y&Bb_6=qe8DG$;m1ytU_v^NP5U2}6 z;LE|Tif7iDI?krDyx-*mVxgXBS{;|Z)jhqr2e18ty|1pt^ChmZsikmB0`00I@GQU- z_F<}((*UekVRk=%0U;rTsHiBApZT@2nGLGoM7(aw6c68gbN;Zq;XN0o40zKCIz(v~ z>LZ#_@~1{W_aar|JD)iERy<}wZnZ6)b1l3$kt0h~q+TY3>YE`-K+}x6m6uV=yZGtg zk6r+4Luum(d5mxIaahYi9$$A7%IM|fHNc`c~K&cr&YS7oVjP14pV z78Z~Ql?!S{$^=6Y@=iEzqQ{4UgfHEp?09Q@Yn~PiBQJa;Nqwf!uIz-vdIC^@Z-5rc zGyGm6QUjcr%8;;#dfTf>FOY^*F2D>enmw;Wg#9-8sK)(8`-!>j>6)$%rLQ4<(TvmO z^0q_)J*fIXH2s**ofFvCU2<#C8$<7h!bdCp`$71-yZ9p-I1lp9jWN`vSvVS*msW-+ zi-JR9?FrqmQkv0%@{3Zjx65>y`>Q^D@RTmp% zK?J9J^LqGC37=ag44%^W^-;GfL%Dd1CE`@V4WTq9VJ$J&ouGilw}80M68bVo@=bF8 zQr33-QN(PWL)-3bedTBR%ayVsl}MZJvsdR=fV3=DQv~r|K>qHJOmU3vfhU2^aVJ$) zvjv$k#>fo`6h(xdej9vh5Cz1%6-ey9b!{MAo%tAyik6nh-$njE-)1VzP6%|9;h^|6 z>%xqNa%M&4p!!VLsHC7{l&@$xwg&WhJd;+=TWm@CjR-1^9O8K6CBrEB5xNslF5=`y z(gT18dvj{a@myL{VYp(AFFCnwu4zRm->fo65BXFfA7P^8x3#%2U{-|;ob^|KVrm8>R z;H$u^WN>9;Ux6_p75nt^GA7Zq8Q}Qhwp2qQ?}?^h5h46TweYXp|IfbyXE*`dbBJMx zmysuH+m;V3=#?L_Qj+?06qCzW4c6MH1De!%|LcQeSroj$br(>WEQoNDC2a^pVY&_Q)WmVB}=Y?;o^=ZC9oHQiAzl&=-gCDIo`HE4n8G{|enhWKwA zm>Z~67e%jPX0mW-<@%L2bOv#vdw=1TkU`r{f)#8BtM5iG5OHf2A@BBe;5>& zLKKUr=X2jL^MPR{ujd-LI)+*=dY2o1KWhGt6C5)6tW#72Ap3!Sz5?qqvC2YHu@**q z=s8Q#>QfX~0usBmNJiy&^Gzl1OCWYBku#pW021QN!rZ<=K-krVNgicS0B!xk)LIh- z)Cf?WcdF5P$G$`RvVLghaGZ!fA;?vtM?Zu}A|Q8gJM9+jQ6E+SVPlD*jBDe zXf9I4_aT7ERt(aNEq7#PxLaBl%{{+~Mmfcngmn_9-^T+SjVL)ZX9e1Ac?SUXsEKCB z(aF>Miu^Y)kd6}q9XoWa}N<0!^pFCb{C450VT#g)R2h7+durW#JYJ48K;F!FBWSoBIV zm9v{2ccwO4gnAC>uY1dH0PgmZk#wEiw|(HTyCcOR`aOES8m1(r`weVkRnAbmxR@bb_HfOtZvUpk#pryE77 zogD5zHA60p4bm{&nXVoWs%oHvz&hkZJGQ)RI|OvElze1h&cn88-r6K((>wbwi~-roPN8Aw2{7RGDzN`or&x%fhCvZ9 zXkG$RgqTY1x&Zb7JzuA3q)2X1++CLS`jbl9(ViCww_5KG!NO*u8{52`0vZ;@BH<4Y zZE9($sMaOEvoxKx9`ma4ExR7Ae~RzF6FjGIJA7Z0dSYa7bwl(S4pvaPavZ~gwM!%K zIh4w)BN`SGq31+qrH7Q51NX7I6ykZ_qjlpl=>|^xl~wnN>RSk%_r*{Vd5R!>1vzh$ z&wwAoYyBCNRt;PM)YdNvU%2);1P0Gm*S-iJ5P2iKzHt-n-V$i<0q~>5g;>6Z5e@m$ zd(AnPN-L3HQ9Q}4=K0MWFWk>QH+|F0kx68FxSQl=j!RTon2PJw0=SvD){?b={ce-1c~Nm~?HaL&*3^8(oiW znVv-w-v>zm>m7=lV)^ng===X%oH|5^1}sh=eAa$AE0N7~v%MH78#_QbLH4iGz|s{X zwnLj&eDS}*LVrJO{{6k=m(Eu6T6=mo-% z-pjxN!-{$IBKz@o(72*+xwBdW;VXDLbli|W-L=eOgp#C_5`9gGm zros#-SAVnLj|Ke}2PnLG@QEAW^wkjn`WOQ>eP1#svf*Ranhx zvY5pq4FS}GoSU0Fn9M0czyv`q7E1BvX8v=j-h*7JPx|yqkRNr&n3I)6MaF%`ju=~A z*w>bNO%FF1;=bCfAf?|h7mXECmJ5rEj24nrl)XSF6Y3q;Yi+WiO+Va*!cYM3OR@Qb zA|Oxc+3RcEtPjMr^5i>v&6tI{rg2#!>CVVmx>G2TI(4nsL~0Q^y|obnP0-IP?NTz6 zr) zh7Dli6A-9pCJx%>`AG&>!AVykT!zJpyn27n0fwlP1c!DOnHc!P&ws;Ff9}5iIr_pn z0l642Uc3rI$?N;0bbALe-j1a5a3-bj&vy_93qv%rgAoSCpk=88J`?0OAqifVtWER3 zaGujzsx8}RSd~rHG<8<4YkZOxSRU+)K1qYS(MGj=0Y{Zz>b)ujSInYVx|XxKrQg!L ze`qVVo%B0CFCm&{&Rfwf=3y^I!2H zvRT<*o6HcSQc3Sk3;pwH?K}vC-chV**Dq7 zV+>_IHmD96`Ukzm#(5{=$ES$a`>(|AC!HcTGe0P?c<^xJzWSBRQ?;)BcPcs@n=?z*+pU#JH&BhN( ziqubuF}fZ^8En_zYdJ-YeP39GhGWtfTk^sIYk%cfU?->(YwnJ=`Rzi2w@80HRrMwq zi0vK%B)Sc3(|u6Rl>y|rxcrFsGy{gd?9KUQjqCP6J?kRq%3u!P)g-R%sw~uMtT=IS zB9D5dHQMD*4{}&Ey;m^io#t0h9ecw^%7fApH6_&=M5wZL$jOI0=+nG1I^@G;22+F& ziEiI{B%z>X06(;X0v{OH?6lg=xRmR*)+b-mi5Bd?qi$&BlDeBZeJuP`E!D!LRJ7M(*z8lz^6ve!PBjd?xsbg6l#kv)SELKyiz!z}+PSHdNO^wl z6&3`x>YbJ{o#HW}SGiIAyr}63t6U{aA*5Fmp^ z<}t|v!?vFmyc1BR0p$jNlT6%VGY3ut?tPyLHMd;-VdQ5DWK}`o(^w0c%0qWy@?_02}bua{hc44B7B;g0vpphBK zhpLYJyf${VH?xy<-#Y!KIW@#@qWx>-YgeT-azCi5s3CpyBa|J_As9DSrT;zGSd{bS zPgxqC4I!`*l@au|&j;;s^!8Fv>POy$7HzO?JT8GWX}F4^rp>j>6VxMRnRR_MoIb$u zl&5AQN%%5&WmZny^f?JBTDmGR7nANwGe=&y$?;VJt40{hxga8|*PLn7(K0W6MA^zw zGXIZ+Ta+3r3~RcE<@>fl@78{QtloN%Lv6SeHdAykDO1)+JcUCyi(O)WT;_^f2SX=w!$9D(E10(S3#CT*w9U_=%dF0q$kW(^8of zIq8-`J2#;hbXBQ^5V^D`Y6}VwrSsD+^jP=r^3egcxbwj>+Qn8qUzIG>D9_#7<^u+c zZ}^!$M&RxpbRlTm_tK?~p0)jP<>Y-W>2FP?(fGP`^vd}2vzGOlCn^+&PBjZfajsSZ1{2Q z)vU*;Y4mkCoZQhb$|oAP4Y`~Jog4@bW@$~opVkp+NGmRo*Pgx|W%tv(D^<*z36IC;OFQk>E$MgW zwG|p_=0U~Fux5FT9%W?=(rr+Vrz{`F%j1&Rw>d@i`OH#)itEksjv=CLnyJd!RJ#Zy15_^8HnekG zN`DeMpMabJmBsOGRI*oA92qXb{q?Os%PO^9)M2Vmc}car?aFlKdhE%T9uMz6ZN5Uy z_)?tcRZ65a^79@R;+hD8fq1(3`Cv}h=x1kMm7r`mb3%1K>mt?Tn0!SkbcOn3o)Qb| zKzc8y+}_QpSEBqSL{*2)sHeLq?T*(l0n3>4d~MC?hSK`n<|3{8-@*sXh1be|Hb1Cj zn#Jyb#-fnsIOda;4s|1?n}t3}LHg5VTm;C6?tKH22ch@!wSd{7!@&I9*FdGO8tGUi zrz3;*RjCZy^38PO;Q&K!-q!tuEp8}+qDRTIwWJ<~@-IuOqbu>hCe%grW*L-G{q_vn zvy1hMUk;46zu94VS7|7!+ys=W*xdv^{+n}XNut+M90uWI^+WIRbq;ZSy6T>J^FqI_ zMNq@`qV5wZvQJ^C4i=9I#@F-imzycGblmJTa^gk_N5a^K;G#boM7~aUDr{+~D8*1= z08<35q@-lea?0svkyRE?<^7!rDAz*p(c(O^PN^vR{CQ#~Knnx+^#H(Lx8fDPZy zY^3%Sk#hdWsU810{_u)eAb}5-d+1&DvUH7C+y}>$4P8lOqiK`j=RTh|qff32 zO$1OS*{cRrOe`x)pz(P-i z0B|PQ6{kDzd1CS{K%l|kQ=#a&pEKNaSTa_AUze7bhrgK0%OpI3PkHeRyZ9ja1@>S| z6G+T*<;^Hy8CH$NqEduAu66uQyTm|n6)fesvU4;l=XjYcrCs@UE(=oK2D5!8#PqcU z-6Ewf;&r>e=0ypKgc@1H$3&z|k{4|+!<&~HSgRPYj#dthKhwd>1WCnOj_c0GAAx)m z?+L^9i;A!9$2-6qMXQaJ6$b71#d4N$86*{;r-e<4nlBxezrn)6x>{RX z>4}aYYcT}bYbH{vn!Ia{r`L27YTrq!kWvAns+8zrXHI;&d&IpUTn8Q7#=0MM9`ju7 zBBJRAq3WZ`Tr8+|W2Z9m!g2meF7C6WcgYS7c1>1PogcY!|JUf29FLasSLrdmM+tCf}m@3Hn`2FQ7Gcoj)k;=j3oSL`aleXcgg&UQTS2UjarH(Hx|tM2kqRu)9h@ zvjThnm6=m0w&X<@0Ao~w-4Ei{`(^^dVtrL@QL`l#v-+f}#WDj>S=cr1t;oloDTFaa zR8jMK&-~kOHX%bea0i^cj{b&+{uLX&QzB?n>Z4ZVg5IK!?}m=2=eqHfQpKmcuC%87 z>45+L3m+^XbYc2ri`E=C_7vb&wE^KMz?yALpYBqn`rK3Vv2Sn#Ss1P)qYjE)l zZ3e)z=lMwDUo>lf|Cdk*eLyP#N?=(J@P7WBd0C>JUmxAl;rj|m4&_t0{qBnUyXCzE zR7R&>Yz#l!G#Q~zRuR`qti1n~?Y zu|8lR?yLRC2VLQ08-UZ{G_)= zpm8>K96Bj$-l?#h#0H#yJ9tovxpE}4)$3kw8xF*(Jl-D)DBM#}^u-|NOz#3xyV-}! zkBl{Ba7~?}gOm<4B&4LiE6jx3{DD&a)76l>-!(`GQ_USuBK@|~$e(b6=C@p2Jy$?U zM-8{PKAfphmjKd6bW)ymY-BWlz-D2Ay^)#~2V21c-U$ep>%#9^i^#%bOpzP;eZ(ZRx%g3n$~(wa|Xyxb2M;2I^PGp@r*upd9Vf zBXUuJYtRbk325r!QDS-d6ibmBssdG@E>~@z1-dzdJM5pn5vi8U<0#TftX(`Ynz25< zBKoP*K|GxTmcE|z;=RW98*rP(p_YywYgr`pG0dXLRO(8dbm7K5Ub^bQ8WOz~dw74@ zIvV+AJPAJBv5F!Rv;=a)-TKTIwsYlfJWM-gv5NxH~{fQU8%6s<1bjxwm>n z_Q^I74-F?}pp5*}xc@kMTB8LjCt#jvMXt$rRlHW;HK4SjoZ3z9wYuocelG1){MiBN zcj6@Gs8{T?_uSh>)MHlX6JC37JxZ8QX6N9WJxLR--++R?sT|^ zV`9~JU50Cj9o_J$tgLL;|EOwZIwr{gS`ZX?L-VWKvOGwulNvtc!&?(F7fWK)3KB|% z@543HymOr{*j8f2!{ilTu%|SvBXg)0B$04#BH}D{>`ZUH zVg(ovN3Xom#d&fww7cSQd@xc)Fg17UI+C?Upp{R1P8^OFB#>@R@g>ej7oI1n^Kkl-3z zf_u<}#@$_lI|O%4a1FuT-Q6Vtg1ft0aQAQW>}T)yoU_k9|6i3#byL-=yVsh%=A8FE z#x*uInM-is6!$PO^kNN;N}DedF!;f2U0h8~6++7SiR{xj1L7pJIZ${61O+h>Bs7GF zgxL50D4k8$ z;va1oe=76+K1==be%PU6Eiykbg6~zsGis6;4MZgW=Kw@sU*crY{|U8=_o(f;(_!=F z*%!xSc?Emf=Kt6xS`lA<5I^|k$E?n9-_~yquS#C4R7L-Z_$T2YK%JeG185WG-<47mUE5TnDZB zn1~pwBb6no_$p6)xv2}-DWxEhaB`{&$KXV78RR#|twRkAW)(EW1+>kmmd*B*+|SPv zo{=3fSyP#k7E~)Zy8wm0rfno10^lRi@!p@wKkboGVW~g=LL--G11O9?Cu?wsZ{I>o zox|xXTSVSr+MP_kJOR$(YY(CMDGq$`jOW2Wd}9H zl@GZip%zn-M=$cs{0c;~XFYxL6^*UDza{^_OvXi#UeNYy2_XmfDTXF|$he{1p~ z0^)_^8;b{@Lev}FH}WY;hES5fbK+yy#F1nV4j@Q^Ftc?6%@gm~Nitv$5w^;Qh z<4>T+kTd+7DD;BB8xqZC7IYI3pUKNKq^0sL+k*}2bs-eHy`1)<+1n=0Ejyt;r2Su~ zl{M`SMI`27;`X-k5n-pDm}`E0W7g8ae>{J!R~_o4da`l84)<`y>PAvj8oKQqa>*rw z4aLFh5WhEtYZR{${)euY6q?v)QSlaIUe%0YGHLdyEGsjv1T!GcCN#4>`f-=6Zr~LG zH&!_JNq|HB7$T-`ubD)+MT;9Ku`kL4q5A$~3SsXlWSwDBf( z4;r;}*ezP}jDNT&Q7f8=dIc60p(hdI2PP>&LEoH59?YFWB_fA)O2~L#{IL@j@AvJA z^{KA*{O>GA4E6?MzZSAWagpym9p6>R$G-BvcaLCskeo)xl-``K#0`|@_KTC1Zl6p` zr@%b;hl1#LE=*W5RLUb9$B!L7Ay{ACG2Z2wo;HgGuZR899T)+IKc#GRUA4^gA0=Gq zIVM_zF}yPnV^^kBcT}9MKk)7@Jg%lQEX+FIa1-3AbRy0aQi0wKomVGzalzo^uh8S} zqk-oJ6K~ieJO94ptcVbGpn5B1CuX82_^%x?g@{s@zrkW?Q$(O@Ty6p?omIds91|Zi z*f?qPem+v9%3P#Y{Q_K%u6o<8Disrmy5aV(pU5>HB!j+`Wto5X-D;^p3Zu=%swa1LZi z=CmaT%lh2HA#o`=lA&Bz?JkG1S;pDE!rPZ}Ujkf!NHtz5t+ajmKIauF zRDr@*Oft;^?Gcf*G~}|w#SPUvyVSIxJPD6AR*nhu1Zy@Uu?n26N6>IP@Kv%(qDqzU za-UPWZ1%9ol}L>#zS=Pr5gZ|Zw8Fy^gZ!-|1dU7w_@mMBqE(~K8TV_18|IeoD;1cu1`3JL_QslE(A%C1 zx69|K3sm3259-#Cz)t*AOg4~tf)h-31sDXV^i4-Tw^RhRz`U3j+jF@yCL9a^1z{6g zVh_^TVh=ml&y%ill>(B>Ki|v`p+(FM>0x``vETFO42nzjL zq+FmtWh@}-vOzYjcp%1%F;h#JF@btwT|JJ?mv3-~X_IkH!fsA<{-&(;{~h^lE&`Ngn~1gG8waLTtfTSQv|z zI6t=)aZ%9`z35iYOZvcTU1(VS2-`{M!$B7I_!rc+{qUFhd2K~F@%3;_&OJ-53U<+y z5qJ8}$)){mOY46Ro(d}IT?o+rbZ{Z04d}LdR8U@vN+q7*D;N5)_vn`*xNy>G%bf~| zHWju>2?PtMSHOun@Fmfl$P-0GrXAt?OKBK*wH0EbjERv<;q$BwW`3Va8&f$^jf~w8 z?jTN!@GXCOZ?Ng#4%(L^5Vasp0*M(OZQVRZBWmfUL&Jxs~xgDN1R6Fq$2`hn>i@Chlu*kemkC5@-Sy0|m+> zX?dYS<9BwVV@!0-lNZAxM6IaS)Ql4?&QUunx%8T-lhDRD71=pkt54m=C<-0g{iRq0 zDu*9rie9l=V^hV)Hb4CTv9tyKg>UNp;THkDr>)|ewGE_BQ||{H34~Gt{S%2N*PT`e zYui?R=f0?JyhPWLOo=hG^ekqjhBl)2=+z1NJ&AZKy*lB|iOFwk;qxcpuNqAnKMHbt z6IhRbjE{nZ;=qR5jHbcZT*;?v-Khg)V-($aeY+_|_w(b*C%%=OU>`DH_GshRNwEM+ z!>FX^0H4|UBxW-jBEWZ{=`ikbQ@|rXO}eVHMzh9~ZSB89$a5gzfzG|*D$5IoSuqY1 zfWOCySsQ-W3MS9a5+9hJ@eS09Q$3=xlGMH>xXO72;xRQ_k21O)%ni-Q;H zP%LCU^WQ(}b@)pQ;UxZ7$8IYkjHWZ_s5_n$={wg#>|Id#IzJK%WUIw|4HB<*QVjx1 zLc&`fs1$za&jM|$TdDG8lgHm)c(ryEwLUOedJZ8v8m7C@P5t3X(7)SdS=lD9_caHS7s=oiC2z6JuqVr ztYTJw9n-ssNTxZ-v-xK4PKMgnBuqzumtdeR3T;0oZkc@@ zcZg7`J>^3PeTJyK%5zH6&t?US6bjHbm4;?xgGDrd8>%~!__w+?;Z+0zvu$YpC^hUF z$|LAs-(IhZk2E!1lbz<^AZ3hITaYE3`zY)cs|xhvPi8lI8fvQ7+LQM8R3 zJtkB{rnLFdRmR3E2$lr9&@$jL@~LUl@`WJ{_2n&7Z~-@=b7yqR{K=VQOmKsN2wSl=`#(mE2gz_c)UZmB*DZtUjfb){q13naqsHb^ zh}@Y}X(f95u%&1h{({d9>ylZjHy+-6!3*MGyeYvj{r(h*i{Lo*N+s@sA0emf9q#{c zXQC6tKirCK55P!{@5Pq|5zY-n3-Wq^OQm?B0gx8^lDkF0DpiF>QK&ya*gNizS(O^K z0_0Dobek?HY4`Q0jl+h<2@@!buQb26P__-g=7mO?ki7F5Fl3VDap~9;e-#!aG;gb? zqp0_!6M0%Pzk((<=N%!;JKWF{l2hRwhc`xT@eZlK(f<)h6Vwey`i_cvPtHjIVyA*O za#vAbmo0l$nh@tCEqd<0U4_9(@6~iP*B%!fklNXI(5>!kwn=UD|5yq#M^D8@_lp@L z`?uSmfb<1ybytA2d`6DT<&4s@egmUOyOjWdWYH}Fme<_f*$d|6B~)Us8TREUO_}{3 z(p%bHoSJM6XUKUg!52NPuhni3DqRM&@^Ixt;ezAm>>YCHK4x?>efJj0jl>6ntkQfj z_uA$}YfAkQbJf$PREZ|;nW)?(YFa^cj*FLczYNX8!vDUZA$qS;=UM%sZ5yWX+BKj@ z`XFQ=yfcpAWJ@_*z0-L=@6)?C1A}%*1!Xp{BN7(m_)avO!#B&Ob*=qlBV&!Y)zpLa zA|(M@AQg#Nn#VnlwOSNYm$y)g;85S%ap%Ir2I-rvYa8=_U0Y|sl3DLVC7yS#aydM{lrN;x%7v!UpBRPrAApmhM+aE0h4@-JJsW zh9IOX^j_rcMtStnEHcRHlnM!s+cP-ve zT~w&`QlyDEXp76`dU=j08-mw;>z?UH<~BGNDPoZ3lTKvxj29Otb8A4GkK+g=tmD zyw?%+Bfz*Ly2TBI|NP6o{Lh>5e{bGU{(|xG2)zez1bG_$iJNSqg!h^MT+97aAoky? z#9pm1pf$i09@Z)ib^pIQ9{w?8L1RiewoRN@hKgI@zuf5l?NfNq6Cuk%DNvh-fxQdx z9exD;KKtlSP_fn;G6OA!?eToH9I50N{y!seZ4C(jHAMgY!M70p;n=p9e7R$R$7|hs zXWJtj?l)D!Sx*$oCyvEFs26vEkIM^-c^KLBxFJ*@DsR1+D0@lzuL+O9{!2j15fmx8~q|9Gn5V@rUN*?An`F6p_Le}`8n0&dv< zKi=04stVYX8-^J5x^%!Bw%l|BvWp*RteEh5#tht=$07aeJpcQHzi{WnvYA333mQ{- zTrt3sp$oJ-14x9!^r4yoWpmzsN)zjYT^{dbxt{PVEC!+{Me+YQ=l}I7f7;WpDCWH? zs0G9I^^nIMR)BVqy@t(F8bifW_<0u$Tz;iUnb(jWXU$8IWZeJ8(dk;R<(mNCxiX2f;+uW);#Gd=z_Qa%mE#Q5)3n|VMD-UmIEmk z%py`!lNK0YTLStroggD*xY5yy8(6C%#Lf*RyFhG!(>~qAzP!AAIdJv4fzT5yc1t|E z<}sCs((?mft=Vom!y{KsAp+ zz2!3)Psktbz~q(VSK)E*@)QXNil!Z)Ar%fz);rK0=6krxTzv9wXK$+AdH3ym>eWfLS^{DmmapG-P<#Ss4r}lb z5k+q8;r?(ber7&*)G3xhsVQ`6$5_~LR4)mx8>;s{7Yr=2a|5WiPqp;W@;p|Sf| z{(HG)XuZo$>XFZN3yM%Ny5`u3DT+Wu1mWG;qANTV8*-y=dxL|fW?w<08GTW+Tg8Qs z#l8V8WsvT3^emF*OAJ)#9ry2tP3-dw@e=iphTkO&Q)0E_>MvVz@2!%zTDFoZxSQ%W zUccw9f+o9=vkS5EH*36u{rMdpuIj3}^Y&Pvi~G;}S>Npd<1^_$-`2*u-c9154cp)l zR~BV_YHD;ttHGlbHpqtE|4v(HEn{RSDT-)(e2WCtS{x>%B=t6fq>v#m5!Sa6nr-b+ zxOpAkX?5t4Ek-fTEyrpj%+KOF2_I1thD&Mv27>CZNjZk@HXE+hnA%UyM#0v zqDI8LowkX^*ZOu1kGYoCP?8NH(=+cw!NOTS{y>3F}wnSrDbjo(QvGD7< zRnZk-ik6-iEx@j0KBpZADZ=(4_KI5+pt?)M0Fz~O4A>l*u=@)DTgFAcOk=+>~tZKy)SgvMvKh<4ZI&5AD# z(wXQnv#t8QGy{tUF-B_eGfG=9Ch2zb!zPzu#1@@#(fJ+`_C@zZcB1EM({)#%A(SHw zn+LbZC~K+Ha@iS$;!B>G=%*g}icX%z*iL*_29Llz$R2GP69NxrO=WXp@ZwW}3U^7xZ*ms9?e zJif8DO`&4v(zVg`X2%mPha;7$%KhA%nxumH+65jess@}n#qKYx4zI8-sLyGp+vPIl zk3=3DzX&Bp;r(?8Pqm=VzVVmzx627Zy=Q^qF+b8ZFFWHrCS@S;+wHPZfj?u3?O(d? zyXq{esbXtaZp=kbtWLeAC?pI|HdVU4GuXTtaJp~AnijhmUr(<>HdokaifBNUPMt!; z9}mV<`-zI|z3r#$mGr53-n7#-9yLFvx#(9uQkwwg-kjUxylmY|_ToBt@$tSlM!1u@ z$SH((lpiSaI_|1+JvF_ZUvm94Bvgo!&oS)_B11Qxutx+-XJp-bH)oXs%MRwvXMCk{ zwVMfY1+4+o7s~d5*s}vB0&(=4=uq@Sa2S>aFEff!$*mN^`*+3-C~K z@%JUq>)50;T^>Hvt2x=vcO9Nfr!_%4?NKaq+1~{7?-&!y^!CM!@Ze9~T)4*up*dM% zl(A-7-aLrGG+$Sler!5WDgmoW1_6HZQ>_|vY`9lgI^^*tOZ?Gjd2SNyX^3QHym>|G zdoP!JQ8jCr&#-KV5kZMv*cM;|sOnLKUEGR3JnIpi(-K(n~yd*vsx(D$0;mIYC zF{-wnml-8-_5^0o1CtO3V9S~PQW7S#S~BpV%#V5nHC&$AAP<`#-Ezy;+nRz1TU+k= zQ|6y60L#}CA8r#S^5irup2JNpSXxU)2hqKMxO|>BpOZ4LsiNpXK)v}0XLiUNAcCl& z;QPvi$P*P?!d5cv3Xu07HNW-Nl^4w8{h%2cWizQ5l5J4j*r4<{VVsVox7Q598O_jL zAuOvB^R4zh*0R>%_?yjDcaOjd*1 zzeC`-x1Mlm;woe=Lbn$vTaU|oZyKsvSi$N|%-D6z8d9kCMKcZIN4{{Jc&>opTEmcR z1}6D!Qm>MKaGCAVTsN3A^CYd_A=BJeg`fA45k#!{x+=6x{A z&`e^+c#CftOwV5O^GADrz|paDqXZ)78@D{IPv&1{-iB^i?RnYR=Q%2DNy=B)oX@8Z zNg~tIpDKn&@vD4fwBBl*kaaFOiK66jh9VhBdw}54&%EM3p^`}}Q7@7B#Tdw&JS;7^ zqqKKdGT-5n{Ks==hC(0-wwILnGF{s18dqbB@-qE;7l(xY4e80e@|?%Bx;=a%njw26 z8`>yaTXUD-pn&`QIpuqT|Py`x9#_1V7O*3H6 zP`y0_Sa@b8N%yYjyV-$;#JXaQpC+hd=moCB`J61YuHP&b9vt^I-_+YuI^PcL(hVKp z={LKGP&pqvi)0ZMe@6KIP)o1+r@8FT?Fa88;h~2mcdwyGSnmMcC$gkF@uej1-ao!l zW}&o7Cf!4}Vv?tBcMABtS%Sp4ebHLpL`OXb@mJ>IFZ7Od$Jm>LzRmPCIk6sW!WJ~{ zbvL%2sRcF#xoK2vTWQ_TbAp8c?OnpTKd?aW9ctYNuqE*i1QFd4k9_Bvd_DDbzF{>t z=$Y~(gyTIA148)2>g~hhA^pUg)f8j}>Vcb?QmDY?lg=!z6Zoavcb^dJOC+pPi>qiK zu2s)%18Mk5UUGYLEh>ehh$S5vHdcBMHK(f4RYnp{Owr$B+Vn=>EVtYAowD#9Uv_#d zHV8B`VC|kKl^b1l-`l$J8LbQd>1s|fqoj2@Rwv?Gs?@GCceg+y5-@jS;zUY~S6whz zh)Qj=uyId#Kqs1_ktXl>>FIf~-frPdPBaxZ!PUHV)9cvw@N@BA!+DyScsN3#fne~-aW=CcV1JyTylLD8YXb0$jQu>kqg6NqiP*|ipD zu261I=t<3DwzuRYkc@qb4*Hx$Ug;d5?xKvL^+)Z=~ia5NE2QeQdmQJtsB(rSPy} ztMI5uSfqPQd`Ns7>9${})Irce&*s*qTq>$HZGLxS)m5uSvrm5NI>mipD84Z9vLZGu zt)pp_d?c0L%DvFik8V;SujIJY{Ju1pw!L!q{#r2McYd&MiD{0NdR0Ye;eghZ==j7N zPZ#?!ysX4CN8Xm9+o8if9b1&^9gPZv1~Pa!T?4-tYv~hQP&;kJx)HLs`Yp0UvmUMZ z8VK0C9RbC?nnv4XeuYh1kNXiHjEJjFyCxh|b6Yu7nj0!zHZL-M)c=MMJGU_AC5vTk zvx{r$qZ$_K_X`I$FzeRuj?({paixp;nQV%RgnWueTYDR;CzV$iF`%sY12YyA3S%qV z_FmG;s^p@e1J6K$rWb-d3u>M;G_eEkRXWb$^gu8}{d4ygWG5?2_sI^GMbvCys8!SN z(Ll45sKP(yw9D$&ad4=qc`rb;>i}H-7MfC0VUvjpe!4$^s>IcXZKPe_STG)<{w67^ zDf?BAFuU@`_8wAUz3ea2h84RKl-=GasCp)lG2kDlBINZNbW5DM|y)v)}nbo@Q-yD&h%gYFjzKV_CdFYO8zBXVqU-j$+kP`{-3`?%8P&qv;e&{70H5KCrw0HDWSQef*?J{6S+-sGM})+mlk>NGM(XPr6D4S z^JtPHcEmBpav0fMayH)@`7hejFBKf?2d>IcAD^MwYMq1Gl(>q7e#s77?ZGgdnYD&0{AH$Vmt$-H_e?9^ z&~-T_#jvCb4VK!TH(qUVYrbDwIb)RLgMAjW4K3MSlkh1SDf;!J37ItgTJb`1hGZ|L z%?xNV4-%FRnxM~2KWp^=fN*}xDbfE;8*~(?>THsPniDdY-m%>Dh->w6>}cMWvI8Qx zp^`7hxB16(loqY$hgnR*{z8aBv_T^8AZfx_0Yp+0!Y;SC*rYo>*3VPgot+J>vie?0pW+mmcJBZ!n=QAp^v#Air@a|of*Ew@ruFA6 zM1IBJ^0yxG~HoKX(TOK}P&{t>NW8YzP)ME{oGS>Z7mB@NiV?xSyg zl0E@?E6pJ179MTJ#N7Ocl-+TaIX5tke^vSPxf7!D7K^DQH?wEZ6Abn$MbytEnhRv3 z^0iNikU(fp4-%N+W~e}5u0UG`f3|D$531+MePQUZnxDSqN~f0K2e{yCxOa7Tn-%hW zyWD*@=#6TwIg!aYrFhLiR3&YqyC%Up7Rw}o&Tj+{J)w7q?LJ;`@>xlb`Br{c7Tf5q zxK=(IW(l@;Y4t5$vDky0{Lgwv>)G!(26W$L zzf;VF2vsOz#u&d&Qu(9c-ijHe@WVbkTS9Uj)kJ031l#O^%tXV^zN&TPMdJC?X|+&; z=evSi4i9n&ZgazjoK?q47&+9C)TkScEyGUf{HPgBX2ViXIU&ebROX`Vve`sjUL!f( z0R(2bAc-X1&7|_Uvr;;e@rUs6Nv$y)OQj@HPoA|aO6J?HKP}_pE4ixB$2o)8OP9er zHWCE%hI5jSN8~yV0kD-1-Z?Zrx!ckpc9x76kjUqOV!ApASyomegC{gDAQjly%_VtV zyygHLG+#1~2S|u%5u{U%#bNr}3u?0fYqlKJRgGnHQGt$eT$Y7nM~U9&K$Idm%gW}O z3OY9+^$qvYGrTSSRuKKlY9X=N7Maplp~=F9!Y9}i-DLkok^C{Him5crZYg$KW@l$% zyx$1QZl=P3^<5oPO>X5C6%nbO3gI5VLoY}N+C93SYJK?TPv9cbw+EDoFXEmRIGTin zg!}@@&Dr_aPY|#!orr_u6bWMCPExy>EjQHPc^BOaVDHlwI!O@y*GpM*z zcI#Z&aeX4f-!ZPFdVMPv68epU+nHgu+5{8XW0aY{>)CfO?P}>)hnxB03pT!1@6g2k9C`!8q7;H?jw^I`j;b+LC2Rf zBDN)-hU8+6ll4*Mr*Wj&fB+a(^N^fA%d#8NTDR3ELu&tNHD445zk#V$9zYG)7 zeGAXau7(_D#W$F1qx)Sso!HE4%4R(&KZLR^-uH$?iRWcvIgI_S=ef~Zl+vp|s5|oN z{o_;-Eu&-Ne#fX$B6!Wpv3v$0N&$R#Kb?>S8ia3und>O)BnE~()4Qo1HpFMwsysUS zvE+nWzJV3DCf%o=wPjJq& z_ri{<6tW3GGw(RJrgDk70MP69v*!l@@j0w)mTEH}6mSb99hsn{KKlVh=?Sc!WZi*i zA7)l>*lQgF0=QiVIP$BR%%NXy-pF#Wh@>_*TP?NBsvsgInzrOycTBoB5RUu5xSX6~*pMQKHGQJX3ge z*Yc>;7A{@)EA5~0aKm3t%67Gbi6@$y z*Wh+JeB;Rwss3fFUWNC4^7E%Hqyy%)7%!9n#pe1AEC$o=OZHvWIV;^+7`qVW3!fZ! zb9D(*&3g?sF3dx5c_D@1_{A`0PNP|Sbt|>IA2=-X$+YBWr&^fDGPQNp9{xf(_*-(wLH^!ES- zWz9)4%0sX$D~;dL`YMgRX1D%3^1B9Zi5_VB%pOI9wOr~Xtk$GBgQz@k4 z(L_N(nXStioD(pZVe^14qFxPFISt@&Cbx3+rR6ai0#!V zSRM%xnXZyT>==&Lvj&Uf)U7UKlIQkEwzLt%3o`}u>TE+k+~TRs-I8OejLW=aRMu3# z7a!1ss?Z9ZLvc2853BA+_;z!GA-qRA6n z1zGitq`%cj8uccwE>U>78720?`yK;}Rz#X5Aq7+q+@iNeI)UP-!dW8yWVIkuSLp0=4XsOxgBjfc&sg`TGl?scpf)s5 z*;vo#RH(XMd(v#n+Z|*~#Hd4WJl?VeZ*57;Jaj~KBWK^am%oHR6AG3RpC>T7;nWLX z@)MRimK{(;Z@hET+v*V9J(m8(r*>XWvRcmGctE;F9Am3hJf6MQiRcY-YL z>_DE;{dFLQ9S}~M!v2m3Y>3}~pxe>d(E*yv;fy{54?IVn{tpf;N4ILFaW{5TG=ju& zolT*9g)h?@bPC8$PEubMmuZ}_?R=d2F<=Cx5I$*#$wntu^|dRg zhXabo<;=QI6iy9&?5E$t2bbwGr(<0UJRjg~DRc;Bi2u+v;=+vr2 zv^7tup0@=x(DM#{9$=Mvchs2yvly57fHZfdw@}l?wxaTD-s$D9Ddk|}O5{AM!HjAI0T)yh_v1_dR0U&6Toqd4%uy~(^h?|1*S>wy47 z`dn=;n#!AXw<8$EesiXz21m;Xm+38I?R*aLN6p}EjB_@IiBvq)nAG~1TZ*A#)==$FALw%7eU!vna*=@(AgKweMcop{5*ISKEgKGt<6Wh1*6W* zOL)pgo>lPM5R)9bFS*&W1Ke)y%UQ3Knw^h{vK%xssFhT6Z7Lo!)$RkQO3i~gq86O+ znbD3<8PA*M7G669{%+8+cR6CE7N*}zMf1~{pGH@k(sNH<%jI#PIH)6fBQRkLSGN#!|3U45~&ke3snF>h9~D9-1CINX@LhewPqjqe9V}71$_D0VP-Vy46=Z%OR8W(YQ`go?encnqCFpL`+5zqU<0wmP4gm2vbl4p!3 z3uc?-_-F54*h49h^L{U7cx0Dl&U8Sfbq zDH8rvdf%yIF(rj)68^Kr4v~7Y=uuAPT>B3zCD_m9_3@NcSdfn8~K3MjP5?9&s z&pf?G9*DOd)sN9vgpuQ)(7MRPSdhuDI~UwjXDW+1b z$*ua*)tq)7s<1Oit03`e#zOkzrQxngE8>iQrgnS9PnzR1zu5A3p3J>8eUTVZLAJBW zVYm&N1QCE}R@>H{sr~cLH<;1&aVP?NBePg`d#Z^&r-}=miNHj@oN@wxyZY2$Z{>jJ zN@s>wUthTJOFONI)ca0=>#`Oq{Rm|e1d-zTf?Z}#!{BLwh_o5#CG*WAIm6oAvv9@o z+ak5?TwQzI2ZlGL?=^yGU5hDcf+3&W5QXf5f3g0&&7bMy$+KQ$+Hy!mw3AC%&wYem zyg`hs6$-snl*_hXvW(Pe|F)JE29;V_5-meKiu#LU$de z`tkV$tW&^jH@)_oyA~vsM%>@@C1KPA)1t4hGgcATVPQjIaD3XXcatAC?d~Uw|S{ zZo-Q{)bR1>Vzr`C^TDL3q6OjACTw^rL)yH8(;9Ww@1BA^EsQ`)lSHC{xPmEmu)2h1lgK^_vN5(0L&PPYRIVAu3F3F?g?-2ECi`e-92$LMB0NvqecO zy@}|PQQ9vl$0X9AaW>toi>*QbJzRe;NkQ*kZTG}tZO`A@?g@f|zaRvz++r^0BSDdg z3R#5o+Pjq{D>0X8SsoFsGG*zsaG$;o-_+=-g?Zh5^9eOwLOM}SC0|2Fl862nyH5Ng zCdO2-r*DeC_11WHVL6YB(GotnVd4?o0E9!)J8NYf6D~>!@QvN;c z2dnNo((g{&@#l{f0>p#s;f5hzc3cD#l{Z5wFP54h=3G21bdN1i2&)SBif@#!3#FsB zxv(9_`=>L{RS7roQ(*KZR$Ub&rn_4U5#ceR4A84N&n(_(~`np_a@jK-MNp$i@1n>sjGP&u2%>8o-MdNF?D+t zTnO2-MCTwzQO@R1IDOh#MS5+R0oTK3W%VCAAF&r@AsURDSd$pzo~TsRXo-7JFctG} ziyQ2D$2Ewl(8j0TQ3+f@Hpi13NKN^RtPcn|V$;{xH?9}c&uRSfiSKIp`BMWYA+^c- za$Ipy1YOX)`A&z|(lF_(3Jqi6B!U7l$jAhjLSyVyzw4&;QFGuv zW0&Gn9?WAP9Qo;92tJn*uIUL7uz5+K(nV%G!9_t_`(9=5c}(r-o;x!Nx#tlDt!Zw`K?ig1*H*PqOP@}O)lqc!!d{`QKC?fsa)bE2uPcM zkP*ul)#K3Rr%U)b0E0;AWTP`cGN>JHJ1#m42k$m*#Hd$UUS~K5HUmNh>zY>(vI_`A zFhe8i!Ac+H(VDt^V(^zT53?)W_W#h6l?_!iMCG^<8wI|{W;mZUHf(i%xF7p-%Ar?( zbbw7C52hkh6rti(6ZOL2&raYt{RWT>rto7|$JB;NK<+~7qS}=zDJ^YvRH3S~h8kfO zOOQ>6>FPTaCW~YrPVg-x6NUADXwqc4PtbJsdOT-^GfwWPBhEwk9WJ}`f|9%)b2Kv)LkmXV&k(y}U;>70MUePTjiz0NDZe~! z{y8+tdon$;UJX*9X zc5^Ipsh2p-ulJCwHRZ}+e!wWj8Ez}dHZ13DDF7}!xgEOhxc7=Gt0*Om62dKK*=$ndMX)urwWUje^w|KYz(*_l7?f*V3o<>F)N<_)(D38Ww zXQbt(OSo}h!&`g_;$edW+~Z7EkP(QGW>~{Hbk*zz(5no8cZlA|F7@{5__LBFQ+Vv}N4rT&kK)A2V@@t?H}MPsq242p&T;@cRF zpl3SuiRAaVH2MC7u8Tp9Bal4>UH$L>yx^gk>Qf{z=sc^IiIGt>CO%vIFT%}# z-3ExP0Yn-ujLi1`5LhBGLBHkqHTN(3@b5y^fBnUqgdd8<`h7gwfB0Na=Q;uvLm!TS2lXP&TdybS|M=8CA@}YFCgeZlF+9-~^q;=` z4AN)&|KrV|g#kNzS%rz>-=>j&3ERDi0FS8RrtHIKiL{g9cRz-%kuD|*Wt^`V-XDJ{ zb2#+=jYi%MgsfOmpU1!eO2F~v#I`+dFBvNT;D|5wCwm)t3L{FLFr0=<`>SCjB9w{L zXF-(3`G-S=y*#ieWPhkjLGVpqo5&YE`7;_#n}yGf!!Vy`qlzK$4+@!5|n-{E4gm|FHdakKu)NpANrQXu0|PCPRs?;uBCYe_qOBZDRs?aK z-p_P@53XTknD>gkI2lASvWacLN6YYK6$0j`=O75Nw3+c$EHwuGryjG?XB!4|>V(cG z`2h(o@7~9>@tB~$eM`j0N4T>CbqBjuaiM{24zpen8yT6d*k@Szj+K?c2U1oYb^b*< zEh7V3RaF(}f`#m=3%Vle%M$KGhOpg!s%-Z5c_6!xY~V=2wX zU3gFsaL4F$Dx&ZbYTT0k*;TP-DN z4Ndo>T={`JVCiXnhl4?<5inC}beoLH420Qv+R%TXwOifd8=;`l&8E`Uv$qz4Qojpc z|FM#}xPE-KE`<5?n=(DjaGF*#=QGNTZ8%Mr2czj(8cNHSW=qQWli*DQ1^5s(!uAKECAK}j7mYQW?pTcgK-PVK*}N1$D?pgDA;@d*04 za)PJPbmU_Gdu~-5jAJ!on?v=}I`L!qAWYda53(>ZJC0BopLl*Pl%r+g1JHN6z2TXF zHu;u_N!mWoJNY*(fg2Ire1`4`Sc*&JtX9O9JkS`B12lN|XcPznXJ013l@ULUZR@XX zm)Nb#jUlPcC%;Q6+n$TT98UltDNg-Hfp!JvE#s-OTeDRdX6ezi+g5?UjM`3vzc-2Z zrNhIUQTGMRxwuwzdJE?{GdYy@hw9y>2xD1*v8A=rFN1;KJl@^v{Rk7P z+#+9VsEnID>0SUc{E8h#SE|Gw!?REA>(E6WvOGO`$1zVK;rZ%BUE)SkljlB~AN8?U z5b7XRVHiKl4=*Xw;WilFyVnsb^i4B4?z`U-tzsptRBgHbzW>4`fg7?HiGiOX!Zf=j zzv3p8o+veqEUz|CS zO;SibpLPo&NOVK07Il(@e{MR;N;B6e8jUjK?wNI>9{3m`>_ng@sFu4^4^uoZwQd zs`%(WSdzf)7&;T9aGS`^K-Rq$9g{thR!2bJP>~^C87@YlSzuOmv0V%DR`Wk;8TQrcvxM!7dBiU6gprI~r&$?G?WXY!K?__)a-A1ju#%Dlx zBKYhE8x6%bMz%-mJ_%oqjI;E(DamkDduZU#vB{+N-yf#PQ=rP*Ai@jTI2zcJPgRJ> zdo^%e{st!V>mjE-o5dPeHGB=V=LO|8aTU%v7m?0ta{cY;V)_KOtxtC2mPpM_+nre~ zj+M0@Ocy${+Xb1G(!1QV$WxTYv&#C*QvDpBkI|{Sps3~@1cNLSdr}rfmV@mat9#wA zrCUu;BAbiR0?!0naIi{O? zUknS7KSX)Gfl26wYL;k9#UA$^aS)=w78e2-zJ# z)4<9I6m5eA%J30P$*xW0ig6>C>jF%0=c+u};UD+PbURw&C~J!;@z>XW74l%yUwd&? zm`?7XAln4ymbr-1%CQ{_a%_z1(nA+cHAwuc$`5kOd@c*m3OjGt5&>s!pU29&Bqx0X zLyVrcbNRL_MT`=AazLiWK8bGMiC|Q3bhpT{dNq~jaWg7zkGk|mWs1EVa~+F8H#bKb z^7v~}NMlC;I`q?W24YuGsNA9QAiiFItkyrLsBX85%i&5)YB~4v7s6?^YjxYl*S!0b!D>LcKJqu z`AdE6XfxoEh_ko5rH8=Q$_4jv_MAz=#I>M1Nc`){97&3B)kJMe5{lUW$KGE@RkcNJ zz&IQMDWyb|1_|i~=_4TxN;fFo-FZamPH7YnY3YU|N|$tZcXxgJ829RZz4!MU-(TN& z$2*3DaXhj2+H1`<=QCqHGuqwmq(mFBF2Dq5M~>vRgb5S~usFYfUVi-GwIz+vca83f zT$gx$FCm+?UJ(kvWfIGD!`?h99GRg@Lg75M~Rm_jo z1UZ7TB>qa5mV^wAsN*BM{%$*DA`h{mzj01}lGb(J({<;% zdO$dBYMyLb{w%_>=awe|#$io*H)jdm*0vk3HvS$0fT*&g?1k3ubrN5bJ;^PYE;R*V zTb$2@x9o~H)U1oF28KH)1hyu!&Ehm52-9Tix$}13zh>#SaHp3NAbCzha+ie3fdo4}Os5fxV=So?GmqG~?Uo=iw)f4)lW_0rYq zbin{D)iahk%!4kzmd+BzINSZsQr(_51%C%Fk<4(QKGeI_VJ>z%B+o*K-&NSdKFcwcG;L#r}r;Fw_0!b z$zCSfTl0vOm9=ziJQ^-FAz@5JqtWzSGEGR!thl&193YomSs4?0UFgD-)$y?H%i6?Y zczjK!S(&}dVcB?%-PBnzVTaisr>@bgC2rNdTLFwxVH`u5syz=T1U8i%FRIruHW){I z9^umP7F<5eXi7{B??EmBs-!OBOOUDnOr3rdAY%FCi3q-gX}gi#6{k7PNI`%fe}aQ( z90G1Nt}b20jl^SkKUD8m_tI*rEt+$Cz#7VJkMrc0r6cN81I}0Py&1s6R?{G%TJ};@ zyn~J7K@bSlNDib8EaEf5EtbM~Kv|9GOL5Cvq_z(~75wZOmaUy#Ss=UF5IPDleMd?H zpee!sIF0+`of&Cw&<8dw#vc~(PjP@(=)j1{O6W#0|NUO@Mu1bF9AN$Ll(6EY`Y;zd zUaQUb4=eEVzK?Ly;MSyHb^Pmi!5}hWKt;B0NOJzPh<}(KXgnO^QzLRss=u7-|M3F( zJ}_coLzQB(|F~B-un;^rv#kHVod17{j;@{dr@xo#8U;8t$IAxkLnUQEF^-$_N-3~C zIfYuOMwB4!ma9rj^4`|M=Bc*bH`4;j@nhi^aN#9(!Er<0F@;b0S-WW4n)5F0tUn%L zR&%Yt?C+y)yo+_xQjn*qE61r@1h=UHqUqvO?#K#>QGy=bfX;$Q1)gkPvy<}lW##cv zMV8Y~Al!+8OCDt5d{-mpl*(nvH1b*yqwtBjBE4@Dx=_c8e(FAI+$)0piWg43+F#W4 zdbx_Ift;|PRR3r<61i?6pKG!Sv5H4>0Qv=G2}Hf0xhXn#IVgW^N;9n>-qva>ub>Mc z2tjv?1qAOaJIYRZANw(Z*>_e-%nSrpJMRiFq`Nd^Vy_-djy()noI`=W(LTe!79dgM zA}TcZxVfmb$Z?;)^2j7ZG;6MjEKJXQtu^7W-jha2&Fg}f+3o&Q+tPa!53x25u%OY; zyt~%Sw%*2Dx_z?LEI)Z-PCKHW3&Hw0bBFR}J?DXmh{#DHvDJuzN0P;aM((`>6AVpH zzOQvMn}_xEZM)}(FJ^#-iE|YxV$rUyd^V)ky(gxR6?L}PmjRCLejPY9mohs{=_p0+ zVVQ!uso~7jfsU7c!*VTWd=9q6EFH9Hvc?>@F%B;Ms1(=NPt{~*_W7r00&ex_8Ab5S zgudx|(PUylHG3sckuH-bpJPHXa}$Dk_;M4?A>>q2%S@!TpW#WJemoySrLN{>t#xmQ zNnk3af!J18$DovQmfL>(+K^XyHJ1XiMW9=(10c{JtUyd zuE#Zlg3?4ya&6L{8J;I}fENmm4+Ote!clc}rfj~nRVUh2C|77s&%cEBcabQ^83 z$~LRLMeQQmomPpcUITvO%y(3C{j8yMkwLR|9RT!rDJKTwk07qM973Q}F5-h_ z5}WRLkE5GA=dR0T&2c2^$LM$UnRXLqNa#quR_D1o*w61w&<}2EIk+9^n>oQ-UCgvS za4%P)E!(O^R56hV(H`I_ZOK`OZx3*p}y~*ROi?_JrK93N-he^kWR& z4$$I4NH))~*~)2RntTl6h?;foPQH3kGZw*)y#s+J3WCfAJibJ=acpxtbfqidI4f(R zEhenADQ_c8U9?5r{1&kWJ7WV@F6BHq|GQd5W`BZM`4ZlgHD;zF&y-Rt( zjy>ww5$JX(s*u`8>9T96C*gU51H^A&WnmJ-YveUW%$cX7K^}Lr1kY|^Li!WiD)Un> z0D(O`N)j48oUCaspwzA*P^}+y5SSXOKgc&C8)mtj7_1i~5UjpZYss7N>|yBf5ef9I zQBqc#WAW*9#M^Fi;}yK+q;KAUfxtT8p%+^1b+3qW&2xcvZc=>g>0ZQJx!SrnedFhK zfhlohEPJ3Gr!s%d^4$xcc?g(OF|{3nAG`NN8Z08iM%Bw?XdxPk z>cjRtGbTr;9%Bg8V)w23<75hrt|QB4PCRq2`GQl!<#-$_1NTZz+%A(L)=uIMs>{Q( zJGlXToo_10?rScVmF%o6@E4Tx7LF@bs>o;?l71QDevZtr{ZRgS9hJG*TW&8LwSJCQ zC~xX+Ce$&#Ca=Chl}gabdRx@c{9CB~5g;OK;y%+qKn_$4WN0KDRvORraO$oRZG>nS z-&WA5^iWq^`H6g?uB=|r$Ald1V2-iqhDYa(qjBu49X0pa)(e*{SX`W+-f(hPc~f44w(LU>$6zz79PV@l-}pVe zn-$-=STtLwpS>+%m@vULO>FFoZ$9iak7;RQgKLqFpFA0+iqED$w1bLb1} zI8&Am=RngQAlgmJLEB_qOk*L(va!DMNO_ec0_g@m&I|9^Ng@Tm<>SO^ErZHg&LCVB z`+6u4NVxZwoGNU)kjdqm-A z%9g+{*J-9fda@0~WIGDq2L7w~>6W4r61;mQV|6#ZdRN=#B^T52xX!)3H|HU@H6j%c zmOo!^*{bSYHvroObj-dwGEVgw>eMYZ%=7=?kx@P`+|s1&EWd7soFi+?C<-)y`^T{F4h z@EIs66AlvJ{_0DOfrR+gG%4($0gVM~N^V1;Pwoqf$$ahB4W4jsHi)W&vzEz@w)5L+ zlKUPs^O~{`8Ke}i6@9W2ClVK%u>NJ2iOw9Azy)mC8(}lJY@dS9W4HKf3sb+EqNfmA zr{YicU%~~FdtNcmDgbc7Gf_|qMPr$!s4KbcBxG;?@DUDHz}6mW%wb7$y`SZ7G9Tx1 zpe)`@Eb*}O!EPywHjyxfG~(DpNPn`!)0W{&TIfJ3Ny7tII4=eGFUlBxUukH?cgXG1 zSqHNB4syLIn6ze@lnqXBE(aMIiAX{`aZP%&9omOgf}hSUkAF_}Q4(G5^R732=zSgP zOx$I@g}WwU(lU4YDGVDAj@~YCaD6VAts4xG$M}>bkZBbb7qhnNlo%qNpT1@f*)h|6 z{1`H1jO9N!Uw3hKTab@zdak`)@(wO887Jpm&JwmO=d=enw}ZFQN)@ndSU)>FmMAOn zSBgS{uSl6qJ;}!7XU-<+Hgn=pUM^_))Mq}DyOBUj+n6Qt(L-Q~fkKpWZA-xGXgj0& zs{^9sfcYF^(;!>8)Sd2?6evU14f041Q94FOfcex1$R*kR2nnw zy$*&c1Xpd<-cpIi=~YP+-3+jy?NcMd8BRu!CZ1EPWR|a&Hn)!rRE()TCX!FZ9n993 zFDl4!$Fgm?hn9D!s=59~xiVwE${#4wwsv6zg8yZg?4?cMiK%_0i<)U7X5O|pEg)fC zJ?cSR?E-3L4w`}4V$DrwOJabSokrLuH87_x{aA)JQf(7TWw;2O#b;m`G$kS;3i
&O!_X66*ZC8aKK7fT@x1dfib)zliFV0lt;i|ij=7+v%L!1 ziFvbf9rW_O6d9ixBU1wf=3{B?mt0C%^nAzKy6VYsaW`GCVq;_PzCn+diXtOr5@cgz zD?4lFC#7s^VbLU1=L4Du&}&v=0wn{EsvU)ZEPoP*3b(x1&0}2ukL)WL)C`~jj_$ye z|M2HzV5wfJ^3SST|KZBc15O{lk*M6ie+5GI1n@y|#-HbXBb5BJKkID+IG7JLKl%l$PJQdzYMx zi%WlOFdAQ8Tf0Ch?jo^I8%b$hN2#X$?T-sC4GauwuekCj?}m>lD=H2wJQ()kzH2-y3hwrBxw~NH zy)(Amr`|FRXUl;6h2^Qbx;owTPI~*@C~!XzeJG7Mj=4aN3ZlxNMOgtMpK|e?H#7F` z&WhFV-!tCe@7q~iZ}HajA@V}!a+gtkp@;dk@LIn70<&C|KIGZ0*inCk1$@#*yu#~{ zN*w%b^OFb=wyuB%5bGDzVC>H7ip15%^Pa~etY6`4~Gbs?R zxsh}>k#RjDDJdzGtDL`9@^ER>7Yr!6yQjiemT<3kOkGaeT3f%W8<&iRCnVrE<{R{D zVuh2`0YV;bW^Hd@N{3f~_8bI|PvxBUMny(~s&0R63<3&afX+zb($eUKz9nT3X=!r; z&LqXfy)SwjgTUI_#DPIUrv&u1jx|H-F1s=D@!1}6Zr*%1PRxML0UHu?Od{{8AxPv< z0i9b22~B?%o^SCGc0OR`Z}lWm{#$Px;J56Uu>r?i zPzdMbRYI(G-^82$?nm5r7UR-?PRbet3>!%gz=! z^$0@e!~JV<&?_cjb!Ka}M{R*nB$c$TQ?I~<13+xNvD2=ef1*2VK5>$PqlT_BDFoCS@6_yJrW%8=xp4h3EoVogJ%ab80(Iw`=dyN8LxJ#P`}2 zY;>ilHF@gLTS7G;8ZEaA2D^2&T7ls7VaL^L!Q!}~oViY)>G|G}p1EF6r+FYqKRkAK z3(c(rlGT9Vs6Rx-vMe{#&;ZAkf{MPW?v}_TC79DO0(g^6k|*}1OnC4&c;USSZ-@sv zO*C%*rdD-B;cB3X^P=DQY>n+fh=b&gQ*oux}ovK{7QCIO;vR)j_o% z7?rw##+T#mM9|4?(Ce~k>XpRP-@9x30#KggRewWzRfzGz5cnPJ3*0RUT0-olC-pZXB2Tj8vYKyec4)R}BsezwmTS1RVKvP$vZt7TiAtx?90mQ$~ zCD0%YWL&xBQVullNt3`HQL7}t7B&77xX~1B%m$(b3$T^f`OzNs_}2lPI(L-r5ZGAU zNv$!=cOn2NC*Jz!+?zO_v|S#2T#}?_i>b50arc&%vYPMp*2yE|wJBBH(FFLsNsGJ# zcZYjDT6&$sqwC`armCZo%Vh;|UfP)mqci*lnkorqzlOD6!s6IzqquqB;@HmAAE71} zLRK`LXKo!~?qQdQ~CIvry^gQqQoI$kbmafC3OE#HV((E8hZyA(Ass~lqMXeA>PGIrRg&%5k~OL+BEAv))F9$7YS7%W}L=xx2@4$N4(dU`_;NAN;Y5@5j8z5zS@+E?^+Ltd5}mgnC&0s5~5>=*xMt zp)7ofX#D_N@UIQVG&kr_P__2F^dM_1YSsU`1h>@&v>uEpu_Zop}MAau!Q%4g(# zl^ooca|H9A1)w6_oOEBalK)gG$u$4Eg4HPf4IC9(PjP2kLEi? zidl^&YOZKDJJQA)V;t!iz5%qvMl(e~503VklS;DbW^2g+O9&cIlEnvO ztoKaY?1SM}4|yG=5q<4hh8pjpKws52yC>Jl5+RA;^BP$9>5_7eao_wsS>VkA!#OtZ z@L)Ix;TA;f=7dkEO^Wbain-0#pTYo6V;18zqBY#rgs&(V$2I+CY<}hU{Kwriw+9PZ zh!R@9=x)3j)HW(BlxJN%rM9r@%)zHU|5Ses4nyb(OcCvJ6ZH5hEFNNF^04QrQ}yf} zQni4@l+K(Kaw}dO|ImYlMNf^6C8!Q{&2)j)YTnX9wf@WcFTaB~W+1Ej!NpMeTVJ&h zNm6cFl#=GUFdX#|J2i{|>`UW=+&2Sfs&V60#aGChy{AI8>`z>i2qoaoABOeR+UJthCd*V^xy8<~&8{2gp&AS6Y zj_Znd&LwGd$Qyr|)$glDd>^_qz^|usbJ#-43ZBos=zOz4VNlEM0&Yf!_UYv0qjxO_ z1&q{G&Ntq#a|W9IDq=4~y%|*I=` zA_o<0+75uH8i)9BGa*cjN~+>`22apohCz}207$eqRZoPRn6@~LD3p@)EqX|wAtwzi zJ^G@qXRqQ(43wHAV9-R^&1=ud;jXxe-#Sc4V?MORKv#w@!-mI4CYUS%ScW3Ipi7tF zY!gnWmvP#r*uE-70NV4C5pKLQfPA>^X>Wf1r!1MzYxugp1G zK|*|?OmLU4Q8zQ!Y3!*yKBH*rmlWek0o=^Y!HE3q?Cg(fr;ip)Fb>SoU?(W-(@Am} z4gSgmeh(%=z&OZXYtF$cL8rgk>3@PS?`yyZIO_@L{^tij^HeZVrvYsH<7z!shJSYy z;I9Fn=Y?pq-1&)-9q2cF!=zU8C(Z%Xm^gxwV^FD59w^!vNN{QyX;0OhBvq0h^H zm7IUPzMuwBv6pY>A1nC3uLP+$Fy>o*M4wInV}?#(JL(!e{y(U8!QrrQa|W;P;C~bY z@1z2%|N1`b%|Fi#Y}z2Clt3C&+zCDRXRq+z9{c-@%m@ODamA*8`wyS}oU_L;LmR$z zPdVt{Jt?RcObm@$3I8!~zy9}SG+<6hCt33D{<~-Q|3ZKuXOvIn)zljh*>$%vk60Cz zm4@5`nvNn}nI1%=e#ikyb_i#SnxlQYCs1cT<3TlYguE^cM0)Z7+FW&(p`VWhlF=P~ z8{_=EBk_b|9plbe5GD~;2rMFD5hrCMG*$`5Qz<;%r0a(6C+f6p1gAzkiwrIg?Z-uy_X#Up@!uMhVPEB?v=?SOj6(f6Ih&JJ1BfE32TtvlW@zy(X+Vg86?JtUr{ zQsY~2s{=-?4SD|lBH8HLXd|p%enHxaKH~c>$ zcrE+uuzIy{NLh(J7JV+`M*A#auxh=m;d=8FbvxJ{E4!D5W4y~VTYfgkDc=(T;L!Y_ zERPl1&~qYsUQ^cMP?LtV`c7uhOQ8jX$DZ$0T}JSk5+=7Fs6n5u=?YWuyUo9& zs760%4X;4{15cict4poa*`mW9p2FMJi-#_W)=hI7KtiiKMia>O#xc~R9MRV7Q+@7` zvv4m8Jy}h1K6|gS+gy&3q~f=^mpeU-D@gQQJnm$82@Wk8L0*jHuc){ag1{RUl?(f$ z%M+&=-a$0T_b#+gVOVF3ukw4O1oH&~5VzX-dS--Pww9T1^;~S0QMf*2@c=rOrK^9@ zx9F!bJ3q3=(8~8M zh$)Z2di#bBJ^N6G(=zBaVfpwydRIVOyt)PK!sjKqqfcSef-DCD>TE@|UFzR9j}M9n z>1T1s_%938aBxmr&EXxXmrh5 z6K8^_X1Dq4_t;0;W8_w5+CIc7f86Z1DzWA)a}p%qZbSlX+RsKhC7H0@j~R=Q(NTF{ zBeh~Q*v;q6l39(b#A2ABL|~fG$H8Tzzm82R%V@*S=sk-dY|Jn@^~Pj&n`T~@YybB9 zH9(L0o)H_K!!|u>CyL@5!kZc9^fFI!0-Y z88|xQ{}dX^r_cB5Kb;3ZD?R{392pNbJOjCsZ|#IJ-y>9iabh3bD?ot((5)=H;S<^S zw#-0<-Gy+ zO`JDEV7B705|AyAsT!=iQ~Fst}lB zaZ&Xo(Q~gLz>dE` zn5kgZI|bXW@nj&QQf?FLX2I&M{P6kC1&~=duq^O|KYdP~;>qfYw6*C$*Z0mW z*L72Ge5UB?cqV_ghn$tXFHByD&g5NRBFQzHb+<7o*?Y(Fv1}G0EB+gmR|{k7RX;;ANKZ1smK0E1)#y_;Tq(&0%i zytL>VhjiU6eeJeQpSsFWwCB1hH2;d#)EPxSGz1r84)GJW|`3kVMceENmkwyw*Q`hk65mKW<^_7JNwgZK8J)!D6_G_;Hva z3nvRyG@d-5aAalRpeYWd(o87pjGla9FLovIh(WkO%^dUla)e`sJ5^Y$A0LrWFc@%r z+qBFG_XtVug?4@`_ zM9i%_%HOh=f8XFcn)-pMm|@t=_krDi)R!R(!Em5;)5 zj}kM?SFnelm3EjJ(P0G$v}Wi zAx$Y)8E6v22K4<192ihkaJk;=U{WnJjTvVL+7o;M$S5})dAa^$fb5KraBT2$cXE&z ziN~F$rlybT`Qj21pNb7SV8kQthl47x7Rvx367@YQ0iY2JLtPC5z^n8YZteE%VOZ&S zJFRzFjdz|VDJ}NO7{2jzpI`7xQVAeeq`Z*#sriE8Ltc)^@vH8834EB!K^(}YcvDKP zG#yRb@j)LM+-{D=X6l)b0Y$2B+#VagnrXhsSB)vT3**T5Hwyyl-7zgfSR)5M%2g4m zHW0Kv%Im)GAhw;?9Uls@3Y5`@`*kCRYTpmcUZAF z$)hWCT8dXGUvt@4?_xft_t#a)%W`&o=j@a+!q1JS`b>@=kzbCV9@+Is_%ydw?UQZV z2ps$>5eOj?Oq#KA?cwV-EK^+Lf+v6Eam?lU!qI5S!)b}PrekT;#V8}~;|5A~IrizL zHQmSlJabR_>2v1}!YaF2pJLnN1lBHdwXFEyZJVH^!Ln9h5p z>@J`aUwor#f2xAKup4W%tYV}0tm%`|klA~wj^fh^VN!L-M4j8>`&Bgk^_==h(z+Xd zJ>@A*67hmV!T=gKA@?#KGNdpmJ*)q-l$4-q0nLNJNNq)D}}|RP62#2C$CT!d-X3aJtD$9(LKHdc9B=em1uJ<`xExk z*b~`eeUR4mbKvOOrYekkrFHp*dWgW`>7K@MP{mPfk%H|)cMzxK^^nN8k8{rD((usu zNP54hq$HhyqBf*e&`^DP(;W9cf8Gmj?jN({4X#CXHpI!;Zn&Nbo}KyRBOJQ;OpM52 zFOErvxVYB+J^y)$*>J&PP-Pe#Na^8@*ARWV^(VRW=LdZ!4fPC>4K3XPUDIg!(4i5; zW96~Ro3YZ7+#2~bg&9Dri8TAlnXE~FnQOGCAxzjK3tcc^&VXu>j<3_RFNeI{;G+bBJit6dedGj<6)7ZLrn;MrZ_IzrL zx+#;4+~(6w>zZ~1`+zqs`9R+SL_r^th09a&5~KGfoK}?$ zyW9(oR$cetkqBe-XqVtvQ-nbGv~u;&d@%@uu9zU*2x z@L3G`GI)iyZ>kjOob;c{rs+B@8@H%Y!yaK>x(S_XPfAMQ`MFfy>j;sW!*k5HhNLDn%F#VgVv(2x zPjs(I8UUSwZ0B1%>IkP8RyXTQAY$|;HagnB;C!@DN3hr7u|rg*_$tr@K9Xz7`GI~+ zwZnSC#l?j?D3|PYj-Wz|G7e;Ot>X4Eat9Z+S_#j(A6-GXpp!`v`g(*4ncx`uWLm52 zS#?3+vbUxE!EWs6kqO-;Ki4v^T2!4fYvh`edK~IaYGys3vZva2Oqvta`lz$r?ucy^ zp3OGP5t2j!eU<7}&vjPgd%Jd7)NLk@L`GAlp}Q=0aj~uUJ@RSvGFV}UFCphuRaHvG zRAw`;h%0vstyEO)1QNNo+EkwEf!qyv)IqZf8DsZ#c*kGbeBvEZZwPs5OSC>s&rTwzZ16H zjs21Yw}pZCc$g`GHdeWl=Ak%m+OTvJ+n~9uQr-2iV(wv0P10od%T&ii&gE$F&pZeK zz5HGS$31*POa`594=Xgqre#4ZG?yADr9&(4dOw7l%60iqwlS?Myp0$k7+mN;#OU)=D={Yn--YLkNzqKk)%vpKNXi082q^TNHH%&`T)8?Rd-3Cj$kC z>P?#Q4A_#?8>|S*ZtJ~>*o<0P6B~^7DM*T~^iplykzN3O07Oe?8Bz@n#L%X zcv$EqXBca<)0DD3QWlcnHqhvH6|l|eEWI85IxdU2qvaNfKx6TW&P*HE z_)CC)`vK5W>Q0@s=Txl=tMv>td>MGpZt)e4EV5d;!@6?SLZnp2V;`i*`1t$9x8n!t z(&G3{J`!B7IfU9DeeTZ-IuR&WiO+TCG_DvzQDW6TW5}Bs8s~Z z7E?@*J~30vmdSL=B{rDSAr!>KO~<~c=o&3Obr3kOhr}+p=bd6C#<}%i+v2LVcb-Bu z?&EDsw^DXVn{vi7k3M#@CjT$jC)XA5k_LtMQ|(PwdDH4g!j=c8T}Ny9_Z4)m&ji3{ zN;4m1#|F|u07tYFaU_6i_M0w4iozO`zaZ^&X|P)>EMaH>ddQ5`rq3OTJg{=$sYm}M}T_C z1zo)&jo4s!gf9Dw3q_M*@`6&(fi8)Q7y>70oe3OL^78Wf6O}f$6E-c}%ER6!iJTUl zavyvWILxT(Tn|~}KCM&k&Vteo*N5>gQDP2Q*jEz`-}e|L zX=F~-M5`h*Z(89rb7wWc#MUygA7h2v#^fB0@FrN!$|X|&z38={%8m8ZAidkTrxUYV z-P9_*W=m;Eq(~S;!*$6gP9w1=hsTI@w_}C{w5TQA#xD}?&&7IrEFN!0^@=$?0(O|c z!jNjLz|8|Y@V9ODSx!LnMw%>f7d7<(wnSlvQRkleCkQH+z|Hn!zw=(Sl}eSx<%Sgv zC{`IIX6=Hg693JO1{$+mIX*@Z@*})QC*m*nTXG^2sv|Xvl29{Lb-J}V^Ps3!zSEYPKB)rh4OM(hCO{ z1H9G{$WXn>Wj@nwIw_g_?P(^*t4S||taQSk2HCm}yy$hS`a=+KDvQ;y^InOb+xk$x z=nns~#9NbriLCZK$&okZ6#et3$@P9=cv-W z*0R7n+hm-04j{=NiYVpoPNxHltHk<9k)AfDWIc4d(xBP5cN{y>q9@lvOiR4CIX<0c3K_B+p~aN5rHe4>&2bXs-Y(sLkR|c zsx0Kmr=w*`2g`Z+5Aff!hpU8nJ`tF>$gZus;#V`xM^Rs6iVxK!{SAk8 zVxIj==BLABULA5(f*n(lLMDk*lCqWpU)o8?vIA(IRn%R;=O_B}d1kg|p@}Rp;*QoT zP&A$4)d`ne7RGNy6lyYh@oYW7l-C`I;9tz>cG(+Q-q&BOK9iS!Lx$4kM!{z07fd`L zA8I-&xL~8NT3O;s_bzeP9%z!wW51${^ps1W4>7`NYl4QU{^Z?+ z=lLdtLZ|ro}}x%k0atGUPKul zKPF*FG(;sfgn{yND}d8%;pGs(XlQ6%i(6w_lcO$Q{IQ<96+pTH+j+B{S~lq+peZ8( zhvgu2;rJ~NaAr2(U~f!17KJ~PWl9Xw{oB{?$iJb_kosYk$(C=KSl35 zdi;{&O$|x|ipa1hFIr_k)zpl-9Bt3d?Vz2FI~*KGh!NAkh_Xa(WCYbyhOGbk^FjuS1OK!R-J?5+`=zF~w z#iauD)ug_E|9;0ufK--U`J<4{`334ea*WA0ol9S9BK?G z;y(@Z5gu^yiNJ(Een>jQ3J*$SBw$Vm`t#7b5s|%lD&A53`P!Az-f$l}kb;-~WoUN5 z#s616Hffu|W+t#)GtVH{UICpWkECGQE01pEp##pkc&hK@Vk4)K)N*hr@g~P)jtOL= zZhYbBHAd^i=00VEafM6Dp~Zmk*4nPoYU6(1dq1BRVVLXpmw=jzc*7ZoFz=HhlVVEf zc5OM>o^lO$^tecMC?=L{HtY@T;TcFT(;1jou*~hBwny9Dgcu_nRM~I$gauBY^~kzO z9=B4D>byyPw#gKUQfE~WbnO*-^492Ud77Aj zj+B_jJ|VN%DFz@o5CAp+G39&8Y@MD)_z$M&Trx=1a)pl@yT)4gwX;=PMtWJ3Ti2678JZr?;BwpKh&&3&&X#cXb^7$^A; z>D+)Fe8Z#vug&T>Xo;?*Qae(UlYwBUI zwvjR64!c$Ra1b=9!*xpOayXZ97sZrP+mSzjJ$JW0kRILH4#*LR!TRl22ro2%RWlF6 z$QOq8gSyrYMF$&}9Ny)U47O!&s6mz-Sku=u(xZK!4Q8?aK3K3h8YW!G=I&buWM}mm z|0f<453*C{1#X)-2G3V3ia_sHR&H`1$%r+`S$e)qcU2VnO;|9k{ovrAItAn8i$J4b zh{;wa>sjboY|0y|bY9K54~#LG$af#*FG@tlk5N3L+t?g7aws1iq%Dl6K{u0)F=ke;?S24BU8 z4{pF`fwWk8+Ba~fVqzTt9^Z_|wjRBttKo(|!m&~R(3&&a@b9ZCoi>cQK`B_`<&3X^ zh~urj7TLHC7#tXTc_`0S6nbDXqHxBsMTa6#pS#{n$}{ZS9$i zT3o~9giND$ZVTBs1>}#zbFDjHf0IJ+Au1+ZzP3(xC>-b+GJCn>G;#0Qz{AiqT2L<< zy18q{fz^l#5Pqf&A>fJ~7#LWQYsC8P0&t)LEM4^&Gv7epKkG+uDF^|aj++jIasH&a zE1v+`_5XhT4!G3VG}TI+>K5>}123EV8fo!b z>DsyG8}z2TX$85cMW8{j0Iq&h!C-Mdb4$=FqMT{rD=X||F3^G7qA5lB_g5(7ytHU7ytIJ?VvHNfic=@)~h*YzE zy}Yc6%?r{nf~h+v4Iw29ZHqk}7`%8$JR_OfV)rL^D_Dyst0HIS-9Evgb76i&FtoUm zreZwL>YU!ygQn}b@>ae3EMAp&i5M9T6>Eci=P&K~LfKm`pM{s5Jue4Q(rGKYE{yMC z=K4?!zk^A3G`%LN+g*m}=%6Q(?K=wXv35N;AUDaQ_U$Dtx>T#n93ijz@hULm^%752 z?WNTMWXehLA@%1=kY|IA<<%~ui_4}4F3)~PL5yvOoKr_UBm2b*9n>;a|aN5d~B+P8I*wn7jqZldiSO zsZS5rC-qa9F{1cU9ji=WLn291hea6j#NPZQ8SufR+w;R*n6-XcQn`r*kr+ZRfBLCs zX3HaqU+k)QrhbOr@&RTA=vtGTL4Sws-a?SY{=$2b=9@4>`Ik+CytzXPhadG>C_P6# z&3Vf!)4s`8EQ_J^Wg3r0@B$Hp#ECJIYq-aH;H_A!044A7B2SJKr4#Z>!fRM0IK8>P z_{c&n={P98T2Tm^3#y)4)`{br{ovNm z2Yn(S<~GSLK#|xBg%)3|G_#E@2RUV|pn%R1V?|Gi?`@i)8~rsq@Cc^rfZ3G^+IKFa zmEXRcW~-?k!#BJI$FB6689D>`4b3`g{r39W^hDYFNX zUP7o8rKY&7f|`b&z3)!DD?egc$4G z&{`8cCf4ZcT0Z!Dw;?M6&_x~c3Y`g+f`q@-roz2TL;S`0?e8Yso25q!xf3YD? z2>`=kCY;=oX9M}-c}e-!j7uymk!PDe?Ah}0tGOG z3@*eKPY~S&QU~gmcqCT7vRuroJIFKDXhymoqv-fy52NyZU3a`cv~l9Uc|A|o5Y z6rk!S;V1dNNe5(hu*SkF4ACfA8i@3qOs_<{)a z7Xk4{gk-21BUpgbs1V!_*QF-MTO@kD1ft*d2w+{uGMtle(nw&t9Y#c#JQ06!lhLCv zpy)$Li@1z*1m3w^(;!X+DX!!(o89u%agp!1^>R+6}py zy|XW>K5QaO?Cfgs^x*v^jF1t)*}^~96G6*;3XMymr^!mO^mH^Z>L^GeG#Sc{I>{q_ z*me#&^E(#Thyfb2d#4ynG#B#ThR*BD*dnAVxz-S4c_Dyrzl zWz&CIhM8*0lp8-~;cuycy_!)2wr~oOI*;$~@BcK|2w<1nf5J7S-rh!lig9%`UcL91 znEiGMOt?kvpI74E&dQ{1CHz6Vp{JLg{f%{Q z$L&DQTd%fk%6zKc<6OFB=XmMl$y&_iSw9bu{H{fM@Z6;|FC*j2f%LRbP*4#6m{c{c zER@)bVY2qRwZWW8?N`ftG~g| z1I4P&+`e?|>j5e45`IjTJ90S%ys;uWFkESD%lou6@ZR2@X%+77<<3dV@yYs4vXmRU z=Y(I6b>-dBHpa^6>fz@qj~y}DLs91H*I&PW{Z26B=>i9URJx0V;HuKGWwJ!VWrlRhGd7krez(GSVLV;Tc8rMKo#G4dq zAj)rOXh=x=#VO8VXc_3eo$)OFNhEMAZ~C6i$T>h@jPQ~eXfGU-K_PIvX}n(-2z*9H zMv~vcXZNotXz}p!p(X*S&c$6_T_+O15O{lD4Ub*}2l;k~4lB0{k@ycUbrfuEi`h;3 z7Sd$;`uaq!Q%Hdb@CMpWfFM$H>uuj=u)vipEPjd*U<&^~{Bwz-QF`^)KgC?J4l0*e z(P}oGAHJd4Fd)Gglg?uihvrH|OA`>Fbget+-hA;+YtQD0P_5o9<|OaBTAz?yxHvT8 zAfwdOQ(qQHlK-40z<&q4X0uvjh97ScdP9$1f~v0an<5*onCTQp=&gMZJ4Ce4?UOY* z^T6Hfy&T~r>s$bS%DHf8i;C2#t&vuM)lF>rX*0!@HGGs7aQa0%`5OTmW0&vjfan{% zfjq)&ovcxL!e8+W+mYz`;gHVT79d&CTIv);>9s9Xb>W|a+pfnj-~AdR00}Id2^1Yv z9cH;1d`{`I9e+FgI!B23uc5<%Y^b65X*OI7_V>PD`~uw;H|Wqr5e7rp1N0SwiK_rU%~9$*{Iyt60D zBKJ&vUx2Rr0w_575{d4zu}`^JqPKRHVGu=BVe}3cfs*VOz2%Jz|1<^h$K=kaL7eX9 zmkVrAB+9%@vbP;PJCn^Blq8xwFa~n+M1n`d;|(#5ksbh0!1^2f!50Ef=;2yxAJY#b z6W-<0iwGBl#`Z#)p~hN6(SmL_uJy=+`x%k2!C!U=9-;Oj0BeHMB~yRQJ`|Zap!Xi2}yLn}iLVu)ib;$aKm8p^tnN1L7M$hF~hHs~Q)(WRsi+pM#so zF|J0pX}^~yqd*|la^NVNlfc8j<`i>~!QOHRbd30GF}zh^`rSc)74OGe;yrnWkpl-= zlEZL=aM?PN3xSaH8}ExGxpF=?g(lvW(Q2X2B>(H@a3Dqi@~$;LPH+1;d73i?B?(eC z@+W1?90#i7u5JzmVN+on1(YsAL1P8~CFCp?+k!x?b|Liq=SY~I& zD~k9&AYD_lNb@|=;c+6Hfo($z#iy2*+K=MDgoVGvfaNt0W4nKVG_WS>N@sUPOLV*7 z(iN!=15M=1m+TUfW7HOj+fj63TcrHOF9c6aFp>(^PY)t6I#!aT^v{~(i|M^GPu$71 z)`-X-sHf6;5FC}}?)~Blz$4HDQptDg$D)!lB1B?i>v#5g80ZfZ($LehdvtepmV(}H zT|BZEnHP8`_*-PlpB`HFEW+cV=xTJk(e(52$uZm8xep&JUM!M=kG=N^_3=Y?SUpAvHo8+;Ll+JqaiH@^l_^| z^e+hb*IgC_T)WzN$C&4jcSaVu3oQPqpZuR6%en{%p5b8-%ZUAPRv@i_sd#YyH1N-l z#Ryz`ZSGV0>Ytt&2;Qmzu8BU0u;Ax!h<$Es8X zqC-J9!`}auu={NcV0~ag{%^5=U)29w>|b`(UsL1%e>?jZb@F4Q|L@(|;diP=JTk{C ztm1&i*ILSXs$?4F<`H>$^s8Ur`e-`OzJ^r_IrzoWtC0guntb!rN}lxf_h+gUYP0IM zVy*(!00YZ&aDUdOAJpLG5W#}y8*K5+&tGngmWW~F;ABcfQL%bmxyN%^rPQ8p)z|_Z zuG<5sQ&v<@y11sJ|DX2W!mG;l2^*GfK@ky$5%K zFzl$oZN%FO`wd2hbJ}O?Yfzhl0>}vH{gXRc4BmMbPee|rtcn|Sc8} zrvl>-o-)dq;FB}(fbv_wd&ec*ceO>di95HH1r*P@Yeu?=BmQ3~&YT%$4d(Y$G%;F# zH9@sREc9x-lczYni%Q^CwL;janMKI|@R~oo=Q6Tm*;OcRV!QkWA{(MeZ3Rx!dKW#shdY0vpQ%ez+rJjMC$Z#xaUPpe$M97^<1 zT#VY40+F)J+!M!V0_enp;tA8suR z>;xh|p9AzJzJk^5hnJM-U7t6YlzX{njtRKq;o(iab8k5`jT#%Rn^ZR4j>_G1IW&a0d&qhTt?=Clrld_NzX#pBCrM|Ao1eHRLl=9PX}%t}n0 zoOKhTwqMSfyE{&eeSXq1uUV#BZqXE3r7(53(0+)Gl~o>^Wmzor7f>}eV3zs#md9=( z2CpbR^RWD^8dl~Oi{UQ+vR6NC7?3T?&9q4#S)Q4uX-EBYhPJB+7+-XtKz)#wl05h4 z?Z+p0dftI#)S%Gb_zV5litQ=7?XB@(R_TORk(Q>WCs!YRnmyL8o-OqCx>R|dORsOP zPum^m$9taz%vv5f`8*Lk%_y(1Nd#lvVie~*-{w%MKjz0m8<}Ppf4r+Aci*a<{KLo3 zOP6q=)lKtMq!)fBZezgblBW^zYv_^v&C1sC!cDySJJah+=sdm=dE>EGjU9uevk5pm zq87{U3#+RT*xl5u6vS#*e%~g!`VMS7K8zLQ`a@&#OHbt(5sp`IeuZD}~D zmP<-%sQby=|E}YI5NM6snF{kyn8T<9qA|)7`?xoG9LkhKyQ7e!j9ldqY@PjQEM)hpWYP&dC zOx21RMD|!C5h?zLUeEs#Deg^erm83=^7V3^;R_z`NNw(xnzJpA)W1=2Vin6>r&;;- z&7lz$RPn6gL)V#qIlBvvSZzEW!Z0M(;H{~Du^0O5BgRO4GTIB;+u(CrCSkL(vTCZ>U+33Z+FNZS@1BMXQ<1!>B%MH{s%pbG(mAr@^SPY)8tC;ey%ZACeNqJ)#2yCykZZ zHeIgRb@N|$8+ZxyQteP-rLz1`0d!aq$hU6$youNM%NI#^YA7}9&4Q&YMRYwtxu#Cl zA_GI)CN?V$$_#Mc@7)>v4_x)-*z(0Ay2!COL914*tn&Sxxf{W%>gvLCkB7ZhTP*D^ zml<6_Qo#s4a!=BKJF@R50HfLk{gtSdwsu(G)vOZ6N2~vVg?gY~gnH|h#aG*Z+N2** z!zdnrC6iMMJHJlWKVHTR2=k#iGv&V^+IJ6M$^__@Bt@C=5L!MIXEUV&dJW+*|JO!- zcbR_FK)HF`I5C6sdzkkCll%+&L+xLi^8Fi@?*n?}f6C&0==**>g>W_gA%HOD--Y}` zLGkM@O)x5(KQ5<||KoOl{*yu*(5v!Ef&W0SB=Ff~=N_*eB3cfwj~ozJWDRovp&Yv> z3+RHOLdXSF1sh59?{Dk>ZT7GI zKfFE^+Du&Le9^)j=R7xT`%2xNpGh&0uf49sgr>fumsotY zqzO3TNXHz5{HkE<4Mq!yMQ$_!&w1!zV9ta87b+Augn5ymNXRuHDy>=`tPm{TNm0a` z@8cBSNs@%5Q(*Z5{oOA&6vsm3kA@Zwdyr58%wYAO%lW7weX!@;H4ZpLzyE_Bol;rG zYr>ksN|*!6iy1pATlB4#Zn9IA|Cs}YM=;_v(Zs|=eF(F>T5#}%F+%sn;^|kZo6?Dz zE{a^>7eNuGZ_OJ-9%bCiB27+@#P--$y>ln;eEUf0$?=^T#PqlpA`0<#1ZR!>wnbw7#*`De>E9V+@llHzJpb|E*cB9Kbx&Qsrw$fA`V;MHnUxS*?AM26!hqxU52ouQBwwe;)4meWs8aW&>Kx)p0GAayue)wz)*OD5|^wkFKGbQDSK0* z6`|nLODde#;8>p}gr=;RxKfWC%bUP$Nb^V!?>~nf`S$hGs$bo>LvkraBAW$eDJxQTV8r^Szbj|H5?{>OWR5jQy?niu|4o!;LGS} zYfqHS<%CmeN?yLCs;a)c3KeidiC7RJF)J~+r%FgDzdnYj#PGm%y1L+TL`lr8tW-@* z5&-k8w#kgN7{6QqE;p?b^2O@Qg3QdaizRb>loWZIFY~1_=AwgNd2JJq+Q^jp8(=y~ zrlh8853Y^fgdJHKbZ(0_fwhnC%rTZAUwgp^IhMDTzWb|;&ic747gn1EaTIoz%1llk zz_9ID)o~&}{98r-hx6mOBwzAnpumDQt_9TWfjIs%T^UdCW!Ho0ELv6gpf4ivdYZ;n;9I5q?FK$XYd{f! zO79y3vtIBFb&Lw4ik${usZH+P_@ZiFmLVHVfiQ)#uN8pj{4NVVXGQn0HZ zl7(*$duqZ|ZdU$^y?3OH(t~fXa_E?5CYMU~msB|8Cpw0^cyU1+yKEqlv>hN=gpo+! zyNtj&O($wEWUQ*D_S}A`*i@IlsJIKF#b z)|JfhIEV@!9T7x~o5jtrW8dLT8d;uw#85_h1&P&(P;7 z48m|AI)~GJZCu3xl0^^vOlC2)8)EZstIy)f6nNxYwxj}kIi-xIxCkuL5~0wob_AR` zwMVVeyYs|Vn!N2YUP5R9Zd0NMU|W}nVK!~G3t;#l)~%VL8|hYrr?8AyuOs)>G&Q4k zmnvi^_PXVaH!kJww_7+4`|R3T-d~B6WjYFxd~Fp|7f_a&Moiq?O}hQ%-ky^YSz!Ef zj=$x1B`K`chjJD*-!<(Eu&DR+5lHL=VVlUAB=qM;Zl0=baGbx5ea4sNl>L&3cV7>9 z9_Eds%Ozqg5KENuf=?P4S6zbRef%o;47F$hg<*L7-!k;|^`$qL^KjTIV9cH?_6GIo z*ZR3f`!?bw++y~E=p`QVZ~}HIWj^OHEZN~&C5-CG&}4LQd29mK=qlFGvTnG&*D1ak z#mpFp%&2pSG0x+!0)ZpgUx97>azjM2lRgRg#z5!gO%0789>e$b6PrM`$%)axtL?Eo z-8LP{w>iv}){cv*Z)l%l<_-nekwbm8$noWjJT%rY9-$TohZ#ZgC3ij1XQyF)$7G*y zh!f%V5SbcP10trw(FYIK&{igpxZqD(@Xu@%Ry|IMs-CG;f-dW!KT4j4&cRj%6QMMg z)MJD~w^yf|fZo+jcjLbDCPe?bcaLpg@)8#0KbLbVVXnQcIP#@D!P4EZ1WZhbu=1EQ zJSeH@sV0}P-9Jfq!ca89ci-Kh!lw*%T+EAzmZ#bdx0I`rZ+ZB%z6a#TCvjaKP18A) zc6cpYcV~u@(up+VN?Nuh*hz1p80B}Wlx5gL;RZahcML0l;cj8KFB#ScduBfH!4Bi3 zAvgOx&#bP|IwO9nET{{fK0?B$r_O&WU0g3Sh4|6m3(AcB-0;@U2GiLW$zm8q9!>P- zb*TNSAUnx(l}SZKMc~d@O{HPda72<7At280I;WavekM~njE?ho9R=gAAejXu-ixb* ziL-0(gXsvyZi?a;l_;OJUlM0E3O89v|5%;q+H3GlSuXKdARSg&w_ZR3EpA8jXfyV+ z*q0e8g!QlVUQamq?#AvxIhuvgQJFo*;+gc=RItVtE5!j|)i9~~mOE+s;45UA{#7m| z971Sxf_YTT9}3wGik;F>i}1~Hzx0H|)a||p$Q)yC>BKf%DOVO;IM`!K5R&da6_)yK zUYcnU_r=BuH+hXmzYRJA838DM6YO34ZD!fOh!P^!|LfkY{P-h7*Nva?f8ZrkOdtYb z(Iso^thJdgB?zuk#1sCPo<-OZpTl`CvhHsC<83!}N5eIxLj~I;b^701vtl%0nD|LU ze1@%|u`vwXR3leL91DNx^8GFJ{;7qkDp8x37-6VJYHnir1R=}qAiauvn%BM8 z_3-Y4DZ=-vdq$`KGLG~RGI@0gHeXN0iQ&iZBw5Zug)QZB9Ary?4=qm{ij8Fjr;Y&vc{X1}97 zhN=i)p*gHVLIRgjezGL(Uqnr%KjB+^lz-+O(jUKKIC198nGgbodzEWy{H}=-ZaMEA zM#|MR$;Bop#J)bDoY3^$313?pxem+6!RWh}_dUv<&B}<(YJzgvLr9kCB<>3=T9On5 z-H+b}V);8%ps%=8v#LYV&ijv1D@RH?5!+k6M+#O=JZ9K;IV%3=@$i3f+!X(o8X3Ve z+3)Z;X#*Ev#mRe)LwlL;gd9f^&RnGog|~ZN1@<^QJGW@8l*E<}V_E<~XMyjyccZAl z2H7~(5}OA2yR*#G&6%daCb5J!j{|?vseveqmCLH`r`r-8 zYEzYh8JwWHIQCnYum&borh@3yIXTiNg2|7Vvz@j6j+W3>|O$GaJ4^azQX%@J~l|_q23{1sP&XM4pH(q>XRoQ zW?wTjoGrYowRsJT$Js>6ZMhB8Flr>vOsRf6_v50t&Sqvx(8@JI9W)L;=^5uY37qWg zpM9Yvl#*>gLF~to{?Fx~^6p9iAu(RahlWon6j-=g=HWW};t``yky>=P1ayizUv~{l zma|8`H;cFeG9RrK4kOYBFfrsfDtO2Z?cIF7ZY$f*slE=W#*LaUydTdTjIL(fA02_R z>!f*LZ_iKX_5MjVfhQ9={K1ksR$P4XZXZNNGl@`53vD}2`b&OPtJ^A`P7thHmN|5q zj$ZVmN$Sn8`LzszF!Vj9@2&^e9vFj3m`@V|Dp3`1R-aacZg-!tjc%QiK(BSR)T$U} z8RT2u#1m8y)$fdcAak&ne-%q`8^U^X2+5%w92qkG?*1F6`MkQuWQ8X+#Ah4d-l$Fm z>~K2YY4Vag9PDO%^`)}S1UB{v^sN-S3-O+d#lRFfSc)B`6{>GO&CA6VdfjLHbF6TTKJ%jy0B#@Up%Lz^uDyF%gHS1a;nk@$Q{NSBtb)sZk>=r&tss*C?YxiGKc*_8=g4xOog z9%Ll?5Kjz zCDZNM-T7jeBWRsn>6-Kuk^z4a8yBcDy{b6pC^>&+dB)@sR@nOhxOfHWbdarHU;m)4 zDKRpJY>Rbocqyiy+eZ9h7^Ga|CD+>J(^1cLG|uM>WgpRxepRvE;?PVg*8+VLAVr`N zy)r5x&guxj{w7{Eb{{5s$V$ZIO+1|x1i`eC@8CgsBj=nHs|M$YQd zPR8=Ht;;RK4<<= z$okdcxi~Wqu~~sDaDpy6Pw;62x;oQTCjpp>MsdweqM9OkN%%wxGt{D4dP&utcN07p zv$X83Q0y%p)}zfOHykQbHgJmi}aaiGq76XVLHJKuJzIKu31pun8wHhlDoH z<(ZaxNq|>sfr+bkZ-9E0ENW%CbN%9ix$!55kqJ<-|8)X`faViVPyQ(m{l=9f+YG(Y zVi>8~RYP6xqUhZQZ-^Z29G!wIb`TNnG)?0eGF1-8*UT1HXjy37FTDAgIInm4Jlzy7 zvZrY=fZv;(uWJB;Su*pPX1$@(EU(VW1!%To$@lWZU18B!j+rU(yqH&=n!BfLb<5i} zNUsl3a9!wwY-Wh!C2_MY)9}__vZ(p#=aF-(0l%VJxLPPmG{R=gKDGYC4F@`7*$C1W zZ}i)!pX)Hu*WrJH>`S^!Tz+Dx_4Rz=qKEyG2!j!_{G#*){;dz*M1@n?#vetB>YHuf zi|m}2h@MN^Dw#@Kl+LZvEPHHGRI10H>dU?%y_kow=FY*%d;*2;_ujvATgh*N+pX-)r-lvb%} z)%{AlU4bQ2+C6#h63!uEVQJYf+QPnA@-WNfWUh64k}M{_?GECY z7E(RC@PeiygJ~g0t*3*Tmnv{(091#ewr+QlMYE(eHnP&I(j3}JVP>FYgz0_tM+SE4HFffwM8n}WFbtDMrF8hvLr6EB@`sFugf>C#Rn`6R6<-wffQb+;cdmrbtyS1_WbrDFFoJB69&7 zs1A7%hhbY0na;k5jd{)Z$xYy^qKHgqYZYkUW4wtufQrrvUtf3dGtCYQA5BEE^VM8A z-POtHM4q2Ft9>!>a({E7a3to)9Ke4zUc>k2^e~hxChe}4q{APaC}gc2EgvjCd&U6B zHm-_=4+9ha1Pmb%+Cm7`q=%EZ^P(!B(pAYb!y4H2uDqxGJ=mcD)#qFsag`%Bi z%_2S&sSKG<0(=?OE0pD>k{p!G!M5h=HJmV7imr75sE|qbDNh`=dyC*#s z$U1>>Cjc=nL9ermI1w99FB+Ta4h&>yCJNL+bmGmS92*|aiXr5HLi&umL`+Pq=K0>2 z_7(0@59NxF)Y1a`75 zZu!W*+mI_Vx z(igqM4Ji->B#uO01(d5te%SpYB_PIn47{DEnnRu2q(mKQJhvH8Pg^hqkIVGt$ie>R z;8uNdf7+>8$P-Ah@ULuz(w;1tlFj+gN*kEXeuSOJvOhWhHYDE5007^si-^R@TA zJFLCGtzBt_K0lhDf9&MB^z$}VJCD|Z!tV|TcU8d6TD@5!E zZ&`W*@z#Oc^(kK7ihkC9Eye|{NtQHg7)I;Twa(_6k~AtTQys6g>dLuI#eGE?QWiwD zv$kC;Q8r5A(kB~}*I#Fe%!wz_*7xd4Gtmy>)8i}z;HT^8=mb3Z&UXLwvWNmqr-Lgd z8o$K*1ZOGE9`Q(;9xFoOxjefvG@1!?a&o>*wVk$WkDh+C#Ze)qoF=kvNYCsLj1&H{ zTwudqUG@`3ygnAC^Oq9Co~oUuP*ZCCxpg_dDD>i+B{uFzJMMzkQ-?;oLO?ysng8YN z32Q*l(0=t4UjJVKXw3^hd?8o$w!uN8{Dx4zfODWE%s-lGzn>bbJ>)T$ZP}u0tYHeh z+Rpdl#QX}-;TTo=VDb4+NB{bq^IYJ33S@WyGM2~Psio02G^D-tw3U>sJU3sS(R2QM zM@ty+`oCUISd$FHw(7<*+c|#T+;@XP_Qa){U3-Br&SN%f1&nOdlFz1u^B8^dEYYWa`o5Td0-`*nhp_Z|4C&8LMf896{18xb{1dWhQKTnYelOhcM}fstbg2DjpNya1 zKYAQ4jgRI`%zx-rIxqzh!%tZcec_=S`~P1y-~!8Nl>Xl2!$0|{g4ZoEPrCIVf|ZN> a%LfIz38y#9Y_1)FKPn0u7jxuH@BbeSR2vuo diff --git a/_bookdown_files/02-chapter_of_course_files/figure-html/unnamed-chunk-3-1.png b/_bookdown_files/02-chapter_of_course_files/figure-html/unnamed-chunk-3-1.png deleted file mode 100644 index 0f0641a934a24aeffa41840f79aaaaca3a6e8bf4..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 66074 zcmdqJ2T)X7w=UYKhzXD&NLEAvk)Y&^(nJL$Ba$UZ&N&;9j3PNo&OvepC5c4I83f5W z=k7P!{h#~l-g;H%-E;0=uj;MZdw1{EYt1#+3}bxb8{fe1xvT{474jY*Z!Y*TH2H?S{xvPh^~c1< zw0jK6#7omulxxS#V}|nwtueDTMROHnBO~rYC_clJeO|VRwRHUxwVqP{C9qzA$Cc;ad5W$H`jxO8nmUtE_gb!w)-1H7l%= zhE{^Uv|mVBsY*A-`_Q;5PWVaprboRtQc1aV#LTe#voD7X`?W@k^5HTLp$l9>w0gxO zO5ta}?p-qTJUoqk(_O!G>QSG}aJt3$KHaz~KfrgwLX|j;;7%Yp=Y4wmjI5J$$={?s z?U?T0r}jMEY8ZNNMJeb~BBgbZm633tmDMsd3G-S}D`}Bhv<;2P4~V958>2W)2e)Qh z8?{pl3!ftmJ7bg&bX*k_6p#fT?abxlZabX>%%MEv7LEo61}lSkofdQiM`x(sUIhXv zY3ZLoey~-?L;scM_FP_;nHf9W@%u5mrKF_fD`|o`vPahFk{c%1Iy#;eQ%E6MBHQO| z{3ON2#Z#wDdg!F3rQ>9c)YW4~g6NX{{jY?y-l2W*;>D+g3ni9Qbu0PR`Oe4iN{_Hm z6OL1z-Q63br7Pdw;xuYNUwaIT?bk-bi(~)Jl!{SMKzf<^7?YQjQ~X9N{*<)?8xu+uM%HwA}V zUPWliplx@^OU70Gi;sfsgKH>(GONDbE@$%e`*hE8{VIg-7Rd83xbS zq#Mf#Y$3(OnlWqUmvXv9FeSk!nOLs8m*x2`Goz)Zqz}}r#>~ruD0z$4X>srMJCI)X4+>d2=lI4~ zQdv3t`n0aWSL*30r`aGXeurlX0?rLySOm;BFKt%k%fbIG_9Tsrj9_D9qfrO-Lj~IC z-99^S+Q+8l zV@EfllC!gavRzZDV-%p4yVbxSFH3#L$iIQ~wGcYps#3!nkr)=%Bt)#}RIBx&@!P@_ zaaBG}nJ^TPHp;c0QzGij0Z2Wy%cEgUv z@k(1{Pl*A!?0!4>3a**OQn%6d&>iVaQ_<39=Dl4+aV#ssKb;Hq0wqb%e^q`*%rSat0Tqb z78B%AGkY#m?gta^uac&xr=w06gt;^1|2~%du0u$Jgnm~XuU3KM&oD+om)%RxqSJ4K>)r16+4LV%?`!#Qb^`yoCn=1bKp?(IEmhT7mdmT@IE_miUM-cdyqgg;A_yDo z%VqFNw>%cB9S((0%jFf^3`3JM-H*o#s$|=Ka1>xx0tc3C5{?&7l6l)fEvgs4x7zon zPu;BPl&u`uoWZ?4Y#5L9$6h#IPC80qCsXLio8o7-PyNq!7Q`y=Vs2-KPBxu)pXJ+f zd2_>T$}lVfnMZ^9f%l&u(3971_$1L>QidhqbyNJGmQ>!cWM4|T2$y~J2miSxA|;Eb z85Y`m8U&}lRAM?!%}^OHI+#f%Is~U~84_Sl4PVq|@Fz?aSsIpF>Un5#hPk-y?=K8g ze52x;^2*`p1zdS$4DJlmi0hiWb>5xfj7oWdY z=2ik`59+C-_{_H?#0}oxPxe1VHd&P8QW?X>nz4$Spn^>_Lj>@bHicPp6>Vm{mgP}j z=JMV;wZXbCPmjJ4`u1hX8dm^;Y&cy1=($9X+c|KsF`(i`-RS}Me&hiReH{zE9Ikis z@s$$J9OO{SAO?36;$RxpS)C$E{*v9xgyWzaZh5nj=C&atLa-pHgeHa%) zc;)YN_*v!FoGnCc!vmk4R4^rv{NG(jVhGlrNX6qRNZ>Ue`%pS@v~_U_mz8|za4$jM{Gw&k9U=E;#j$2M>^rrw=7^7zdY>EU&z$#-cnL_ zuATh5+AvOOh9;xu@BIDS1CRfXbs6roK!~_Kyyku~G}^-1;`kw`kLB(i#kI*kt{Dkw z3!`Fw%sb=;A@IK}9lF((PxX{r<|OEZg`323o4nj!i$wK1EU6K_{Cicf;vyCY*;Gxr zY5D<#FX$0H^NHn5#|zLx6aF~?usSW5hiJ85Byw0sIm{bm2^RbBQH!|c_ zsvX7BNa2QCcrdgaxU+{}?`6SABc8KD1NV}TL?dpZdN|<(Vp+@!r-vhq*6r~#&mI+E`p~eA zSt+;j8OWmdS1by(orAR0Zmsu8i>LSAR#n=+fr*Wc0-zx>p3nUj6P*^A6TVE8B!3si ze{Ey`cV-(#8anX*6!5}(T#puX-G7!SRI6R@fkB5ndueHDmNi=~u%z$bzh6{T00ooDYxZx+MqB7`5L{Gx3UgEGe z8qIE01|a}1Kem^X^=R70*el{hLS^OA62)Ho&O%r-xgcq|P&b z^w2h7^!1vZE)g%!)vvAl5Lj%tvrc)U&pLCy20E!fx_VXdtf7vf9J3{^UO1eYnSp_U z5L7mntJARfeWjAFZj$dcN}j^O-G;C?Z{DPv6C!3l=)TTTy}2&z}0|eY;s5i~c&9n*!K0Drw-j$`R6V;etKbwDo?OOSmriu#5<0(Jn zkUfF!S=4v7I^oKZg`KUx&bcbTb>ye2^wUu3*iUKEK?aL*^saFuHIXM0*bk;IA#_BFF}Zxx1Bmv9e5n|`WmNWV zUvIF{J0?Dcc-;d=#M#+dRomTuRy?y>j%9Nyd?F3e^M$+|&-i1f;X8L3jIoG8KBE-K z`WxMMC&YjLY}+k(fON44^wl>qA|)YFQBuNi5A(Y~M-UhfWGhdIUBsiTFdHeV>Z6y5 z2@4GkjgO}?#=ra>Xbu&XB9CXk4^Gb07u>shH!UM07J!9Kr3S#4^#Q)YLE65RXM_dx zl779nxvTA0GVPTL8d@Zfz^*7LDt4ORdx(un9TXhFm7lh$luZ<3riIvEN=gdD1W7-A z3IQ$!jUqh6MSz;X_*e3K!SA-;;Jg@JZ<^OIL3QH9M73kJ21{%wm;3SVvL8$2QzYy; zPWxpgf@q)$7P{hBG;5S|RNlXTZ_nDn#L7zg$)zv7)9`AXj&@`+L(s6_$LDIo>6w+I z5Z^LPs?;?aQ`ps&S!;eI9K$A!LFMOA=n$E+K+Jo1+Ik5%2J)?=1=Wi=0TT$C=koFd z!{4+&x6WQ37oS|&<0{dC7`hI1yaJ>EKG~zx`CZ@{d?oW5w-H`3X0ProPOWqN^en{h zKkO!&>t~rTBMo9pdx-J(B-L0Red@S75CJ->k8+r+OKFoTH56O<=Mj{ zV)<9cB==#?tr;E8yCH>8hp#k;PIj02QY1ne!oVarZcMsQ)w#3QY<}<0QoJFyb#exj z%)9nEaz4A)v0UaI<~I>~alPqzJNZkaWu3=eYb{xy{X5f3j5F{AwAbwONAke3(q-~o z-n5Yapeiri23;pt6xBQ<8Mc);R){`Bdg$LXO2jXTHWIw>w(hmb^&@ShWQ zJKV&+O41)MWI0*mvQk`Bv<{?;p#_GdMR*yKc1!$Y?|mu$KaYCe6<8042b~KkABhw!1@$<357U%A3 zUyC1t?G(wKC3Zn?Vo zo8+A|;VN*>s|A>3Ut8&b+#XJ2GWkV}4d=m(taVp}qx@nF>TPn*WjJTW1+Zc!Zq)Om zzWyNp%E{H)c-gVLkz!x(U>bgi`7v=eqPoeVfM1htN?<*dT9kwNd86E?W+45yUOGJ<8B?UOZvZOEs=l6&fh>*l2P%+77j0h&lUu)!1_(DqJ+nT3e_PlE;Nnbx^f(Y?tn9MEqYve|rhiptP;x}~Ra9-oe*dk6ihUv`5Q zeI#n>=JMP62^YpmI-e&Yhs>oQ3rBWCU zBTUw>(1Px=FVPPsSkxz8UEsXx;4AzZ%d6_rrpU~8p4#Yi@RmuwPW1J!qEqGk$oIN8 zPxg_3=0(B|?MG}MAUl&j+l*3#TX?))_;npSqt<3oUr8Nn<`RsW#?Q9dq=3MW1T+tC z$EB=LdIQ0w_?twqpTy!ReMY!MgXCfRR2X*3CIxosEfWRGUcNg> z8AJaPw%NzY@#^mxo`*qPVFIzFU;AK6VOfc?nXX0qwz{7^M>*!yBe0~~-$l-AZv+B9uQp1edu_g-COuJ&4EG5& zi3sj8z7Q1d7Fr@NekuA+X~1K^zU=A;lCfk#Rt*R7nkBFI zPhGn~-{TQC$>1>G+r4$~udUOHR{5DM|EkjeFz%7HDnPrkFZI~Z=`1v!Y z3Ksc$G#>(~Ugj`)BvBkI()Zkkf6SIYb^;r5D_cp-$L{m(iH`LYA$LB>L_9-otfs5~ zN6eNEcQw{kwrh=R4w2#E(ZB@t+G`>K?XX?Dv-!4rVOCWKD4}bOI&-nKgSz6U`vu@Byv^?g4Rlrc6h7a6azCA>xEcGp_`;wt(-iRlf1#4{oT?59 zmylsr@KkoTIQjJXy-zIIO?c}kmw!i91#i+?^mGE6hC_7&gpEK@r53? zc&#hH)5x6YRhHRV+qV_Dur&H~F21O*&ZL-h@Gxr_9 zxDiJ7(&k8+#KeV;JR!b|{4_y5PqiK1K>{qxGbc=rRRoFYqXM6ULT~Szq0HK2WJWZN zq_je{-BqpR{ly#e#Z&!S3qJWvWL21S0UIfp_b;xzs_=GywE#NVCMUl_-rnGeFC$1v z>F>ZFVL?e1j`-_U9~QBFP#ASo>Ruzw{UVw_T5Wk)i-C0MfYB~@axD{gs$PuT-d)F`g`pat7>%u5$g56sYgmPz z-D?X}2gjtIsdEKH_2`qV>})(rvqUzjwjXBS%b;^!hEHJ35|W2eD}-Y|eu@8f6qous zye54TEZweRSbmo_8IOFG(y30~uAj&L2~yt1Cai3Yf>*`H|Lmvf+ARI@2YkOA>Cn#u z<1C>uqp4?@Y3KyLN(2eX^re2WrT*qsq2=eqY3?{_?jO?0I-^tf7Y8^*G2`;9+29`< z=6_44@b86l|3fi>F$9}5S_K|fiSlmJW)5lQ{G8>*nbs9=+cI<6@K)Kg?>9%MIxHvu zDd7$HdW~0MdbnQo)Pu9Sba#oqjZtO@@W!WKxFe1l{y&j@X)@`%ukXm#8J`pRtXJ%R z^UKQUcK2ycgx`tX9!HvTnfobSDf$-CiHROdBs%4wH18;RWuOm;+2Mj372~1Xz29~P zr;dePx_MN>P7@iaKlE|Ag+Xda&yhpE6@hqXdgc*bE-wq~A~b?-_To=D9N*-qAreGW zB#*%N%K^Xar-Ze4JddqAcUfq~w;197`s&myRJ3KKTocb| zfaCEM@8dGxtlh7%&Ry&{q@CF@VG=&5J+3))lj&tgAavC$l(p1RZ=pG5YG=oknvy5i z8dJ*t?sVrC(gI7-r3>3)168Tpzl$$rcVQQpEfFg{P05 zX8WbiV$Hu@`Pp8=e54zkp7-J|N~d+WlvR`%f#`V`;dbX6{#RBDD+xiwtw$G}qS|rb=w65UVtF-ru=i&0=u_lrh?Ik+ukLM6F-)1T=kz|xu+}Xe0{YYudF;JGhQ^J_$ zn9=%ozneDMcC@X3ukS)H7NX~pzof-eg3AtMZlRi^`rDJTX|s$zkLGS# z^XmvN@#@qNyDkeW3qyYms&yd~gvcf6&okrjwHjlST8Y4cb-^|C8GpMrf|ZumWD0BF z)G~8H&q?fS&>`!|p%g)28bdIJhvr7~RY>8?)?3Q+siBe~Rr-F|+@|~H(nwjt3-d>F zlZn#pC7A~o5Z3pb=2}IFAt4)Ik1j4VuC24J#_YoE-u$=VhfC$qwSyP1g63#?P~96*e{DteGP~U+%WA7Ax(p7{ z4FuvsAFY^SU}Wb9{YDR7h5J>vdC6?FjL0TGe_p`)!bmVpKoKW9Rw!Sln*ylLnK7Dq zRhW!k*f{1_f5O7@giD1q5Lf_Qc<-bbw*UmR@uwxV3%^Ds|%F#RTFAuL_C( zjwtKjW9%?&`DnUKtkCgd3b3?5jRV~Vs)?23m4fcJw#${%W0v)2`(u_v{44KCwHAlJ zzEX7j%gTt@`t2dmqoxI&iNe~*-ripIQWNFp&#gdp6V9ZHnq&1`0?o~)K?t=Vc{nu| z`(jm@lXux&;goEJYmJ5cHSv59L*$g1PepwTrrPu5>1N=0AXJcL`P3Q1u`8{O}kIN6&m>qiJl(7f&%>pL|mj9*UOL1SFc_b z(;v)zVUJ%%fA=n@_LxX96YpHSU?0y~cEXp#niiCM-3!lh?ME#RBZs~h`-IR<=p{XZ zSz+{GYVvr!%lnA>ZypP^CJu|mnkIp1N1#=Rs9*00`*~$;k-&V7bbm{ARQyyG8PxnB zC3ze@&jeV?+w(rHi@|%3MX&ui=@VA;OTmnc$~EwYx3rpfIX4sC_oH=hs%lt^-Kq@$ z<=Jw)<79L~04;%%d@Vj-rD@aTl2}y zx4T-lY~}6fdiLfGd`}Masq<9Ef%>d*jORB2^PQH>5B{1l^R8>}mv^;3%fI*?yHjMH z%Wy1%=?FoSofkfG7wi1R{fBGJC6=(-e+juvSyjI! zUc5T_Py2i62Z+YC{mw+v`R6Uao|rerY@p?*SRSp-;x{i`tuG%=%|rL{s1Hv|@3@=> z%uAGG_#`_2D53WrAg}?Z8Awb{dXpH3(?e^5%N{h<4M#hbm6ed%U{c9txPSj=SJyho z=3p%XQMjzxG-pG_r_?RaN$|^lf~L*-$GhZ}D&InnWbv`GyH4J}_YNI5LTz2%QU2s{ zV>_%v5d5k?`YF6+^qV8ya*{#(?a4VhE3e}*s-Km&*R=>{cbSi970@jHBHz~I>$V@= zBCdlc7u8Rw!H8z|FbR*Bed+~p3-o1 zfqebg8VfVC`6<5?A~!L!%}V)wE^n=SY=-j6M+2#`dTXBtHr3xx=hBHu-i?2W6LM-8 z>l(%Pk`z`Up1DG)Ss+lTy)w+mcP4^_xR1X-pMis>3=IlG&=O`BGotXJpYhlG%^&!@*)T+yQWPlR5W2>2TqRO6v z+d|{xF)XV|kn}-@YaMh7dzTJCC@G3BSmL}hUunAl`K7!;jG{79%f)yiX)y1lQSnHH zwY%s!`i-w1$t{IrqVu+C_OVtTcT8e?y5kgAe|sDP=)Y}h#{BZdS&Ck&5re_)U~aym zmt)K*F6PHW_KDr}v|O>B7Udm@#ZH83=zr;0r#A2@pV;wGnxs%4`4xklLRr~h_7z4H zS>X=>sJM%%Nk$(d?P-Cx{R_hs+l6>X74+$9ad+L}OhBG`DTZ$s-Gl569QGM5phoLQw25Ryp3QmzT-*P z)nnA!UPzD-GG%&5?LbQ_J+`llvMH@Qf2wVX%NG#}?5y5ah=08!jk##y$tlwd=DzCq zI|9qeFt-UJUe7>~DkH-QDi2~FqVTnP#_aAmIa%4V<6!xK`;3g%eDTJ{<6BJ^Zrr%x z9Y6{>oI4XA2oJ*vrgb&Ho{Z#WW~zPFYX;rgGK7>na>8dN1pkurT{y`15*wk{qNovB zn`F-g_y+%%9~i^pGgQ?yv`4e zSU=pD@)3@CwcRQ0Vkm06cE>Z-VV!(k&7peMhvF`8+I-0}+8ahGn5p1u{nx0pKP9Ew z&gP;7nt~hW#(G@VKVhitxY}Xd&FJW8US3}65To{pQT)y;lo8#Xo#wAcKxrcuKvKKg z`xNpe*_@`}D`*K%@%BXmi-TfcS1r*?ecBY8WH6p*_&3 z?k;Yf6&JJyQ5^haD(FZMJehRe5ZtKQIoIBmDD3(B`}?e{EMfFfw~Vudh~igmj_)47 zE}`?O)(gtXwF*Y_HAj8Xp?}2h>Q_%}#f~+tZ7*1I1XC%;%ciN)lsYE1aHi{*JDeVq zq||hPI>!3AB=l33!_vLBwjc^-!z@8p{*44(N+uf`zUdvOUVeKOe*4QFCKv|5DngI1 zLECvQ9JV#cKo=6xrwjf4{egjjMD((YRqJ+q7-Ru(Ls_*en!R;pPd55d8efB46q_!j z`QSx8roC}``}+ENd(X9l!fC>ONHg!{OVkq$5oD?atK$k6ZD?3n|BqcQX5ImzZ3&sa za9@b@m$(bF!~DA&p*u6$&v2Y#JNJiQg}86G(dVe-@wx22215e! zPnX?A>$TFM`9FUyXsgJFL3$sTg3D}crWy3hj7&@`kXHjad;thMBisfsQ~*Lv23jMK zCpK7xnQ9Hj)dF#FjY}jI=U;(pOvYbb2O^n%@cdsvep5LBB%=gpLy((;vUsxI)AL{7 zz|3i*wj&{Z$5qYWbp0Qc7uCoXn33plZ0ULO(%Ra3uL(RpXhNNiw&w~!rECzX$z?j2 z3x+agm`)F)LV}~?A)Eq~qYBY>?eTIBYt>>hNK|QrJ>1Z!GguE@pYBKqo$dBPe=R1e zI1M-cl-kVwY%o3t1`4JKgkU`X0;oiP78bgr9*QFQ?3Z`j=o9<-I#94FF_`WM0yU4r z&EFJ|5r2y>#Fiz~bbSEQ9&p?u5QNr2>5QBNA#D|X8z3n`*vybEtn_t*O!mpa26vKJ z)6IXqqYhYn+TJA54*@JynMpBipXqTt)QSyd!bbgjetf#~^zRYLs%_!Gx&?d><`MT8 zn3->U0wKnQEHXa3MO^N2t43UI*sn@*ZZLnuVq|POUcPYx&fBPyjScV~AlH|b?P?~~ zJ`M3a6&>1hk;uu-bp!Q<4<1!ae0)5lBVYG?d1+Z|f3QA*TYC)d#Io*a-nNSuI-mA5 zl12EaO)kL`ZiS6|%T*`b$>bNch+Q%r#6SlQZKq&5ns00r6DuY=Pe0^2`e7BY?1 zrxU$hXIV9?cB73FUR2o`w+2&*)_wEUvR)f4CA#-4)0rCuNo-I7oYfvoo;1>+*dV+6 zku~xX&$)j__7F194`^f3gTULvc;)8i2EN{caDXYlVg{BtCZ7x`Us;gVPd9i&OrVKW z0r7K=_F1-}wY>ZwKGt>QBbv<4Wg2;u&^nNhLU3ZI2^^;%Z;b40s$!%^{`6kl5`w(*(L5*|5mS z*omiavU77Eu&{7d2V9tQs6L;F2CZYUtJHI-0fEEN@zWPzqR%CIV(JW7FK1U&R1_8U zuCfp+!Sj=@WyQsE4FMNdx;51;E6*y(`n_LlWm;|&UEvrk z(5?rOPrjl1bdn?OBTv-nvLC1?Xt&Gj&+-fnkOhYjl;QZC9nOZBl0a9+ep~teu=Lo8 zonE1P%rQc6`2G8LSVno*(z3Mvipd}38MH@$A*=Wk`t%&yu6lyTefJlYVRn`5pCLPvgg(n&XUR7pqZ3dKJ_h`>f4(eg;nBL%z@1jp|(t|PePFckD+r4rA zXdL`4OqKA-NVf^ekgH|QJNq(;>4S|aVQ>~~eHZ|P_!0poSZMplot2>iVf*F&lZ|@J z2gq+!<%tOxSy(*&e3EDhB(sG9mo0;&5w@6OH8sc9E?`kr<_9?><|ZHWI-VH zc-$hVX5@MSRp5D+GqiH!6Koz#Neoz5MP-Y?zu9tv_MbniwgDd0Y)ca`va+H9rcygi zd&AQoyewOVQ9AWRZ=eUVH^5QX&e}nd%x0tad7RNQa|+)+(As#%pTp(FXr2zwP;YVY z{UyPsk)Mgvx`+{w@mNi-6x3sC!y?1N9;lbR{*^37+Y7+{DQoR6YzAs>i@3feGduRJ zKW67okM~x@^dW}5O3Gs9kCE3RZlPd$?fTMXXtzHIpMEne9$@~_(NO~~OaizAFwd~8 zGZI{~ExhpK(zR6gejkfGp5}3rtja$*^9X4ra_Ys@b`5W&m-%Wvw6bqt_1Fq%XQ^I= zDxN4|PgD%2DTfS{kGZVml~2NdFybf(t|%YrOQxLoQ5d7w8=ts%u#8yOgMJWk~s0v6+DUzIi~1n=|~G{ z6ZKgNQ-$NNDJ_FCE>eCmpOUWe$gPbJtOoZ;#w1xjm^xB!vk(o`s$)s{%xaa5R~&n+ zrR(RW`anoVJzcuHEYyy~-TAEOpdv1txcYu&9$k5OQ;Q)1qUS=k#axKB8-*X2GN#2vw+ z85179*TLS+!SN)uKutyE9<6K!``x=@NLW`7!H$4I05Id}3FTO*W<@QsiNdIYkgBqbx33^+7eyG zsqpG$7PFH!m0jIzm9}42x$9azwQ0H+M}3dWNBr>MYq!y=Gpb>UUE=oWs)aL}x4y2nubWtrXg%T8BgZ^S zK6^vyP*IO@)qC-*`#sv49wI{qzg3I=Q1Gq^vVVQ`v=T+TH1zd%GXgRA5FDtovb`p{ zFA>F|`~95{^=B2}94eyS0GD)rXu-h*1wi2ygntC+DF{_;1Y{5Okkf>_uzmx4$JP?S z5d*kZHT?)Kkl>Cu6epeSgVTZb=x9l`oPPfdXBvV9t=s5xm9?f&<*Xx`9k7H)^9|Jt z&W@kzW-1pFUMeZsjTl5SzbHyIOiN~{Kb>Pm!vy*N0Q3H~M(+*J*ae$bDo5WR?A-#2 zFJ6ogpWWJ8tg>8RK{a1BNp~c8C`)-zyXC(>cfa_UkYFPKy`o*LZnCz-VF#M^joj8} zCeko#!g42zf?Gp%f7*DQ8cr6^6YlCJ*|+q%71XD?qonfE0#hX13IuY@M{a^n?ziY9 z+I<^mNfXH-Ggi@cvZ@H8d?aRSAT!z*qMC7FBCh;?Bh6^nvhGnD6Gzw7es`tC*cm4M zL3t1{Ub&Jrwd&M=y^mTY##4k^o9=_`*_ZPnawP6|UY@ zBp7O4Bl&VOVa-1$WkkWtB6>}eW!t;?oG0+ZzFZ?h@|3^UHKnO(*0Tn>t`m;4l z3@8Q;3mjtIRtp6LbVQ%3H|nZEb?$u(=P^Myz}z3*-SyLNu3A(rrDGt?ngi>LqCen) zh(FyH^*M!ohPypT>)s5&1)S%Pl|3z+w!46|1<|KFP`VRLEm#Fu@?gqS8woUwie}so z+LQ0cKkkI|$44b3Ov3sCkhixnUSb)-bj9gfM4w`kJSn1Y?(Zz-zKmk4ANSqIRIDyPwZXSg*-fq&J@n1f$IXq!^$B zF}m~qh>>ZR0`XkLOAF`atPB&`(^(qyq7@HnH$(Ua3CYDSodPZpZbJz>|K{-iEn1gm z(}G$D@EqO3=u_AQ*>aSbNB%nt-3#7xgvMU*#FMuj1NQE41(Rz2H3d;G37M8QB5`A> zO>%d+^Q|9Z!}BSW$=WiYLhm0fKvMpdE$2N&`u^3|t@0_E`x6o?+m>3EG^mu11}@)p zkGk)Oljn}?m~Xye4|*@3wQ)8`vWF-zpP;OoA6vWN^i)lqa;7NQ^8Nv-vmAtx>->dT z*=OeuZ7mHCCIiaCYLlLSl)t_urq4i2{1{97I@0G@(ajox_dG7s3>#a%q51j+Wcu?j zW(@Bm)+5#@(#=C%H#S0a7|mmTrAissC)WS0mFsdRjN$0!T}YT^<}@4iJsSVfPxR$Q zaQY+sLA05IZT;pQ&KjLd+lwq~u1y@rMcAS@f3qM#rdD=RBt#a#AR zkKoS$K#;_-yZQL-#pae4j8^?*e+{b7N{WkDfSw2WF~k{K_2Uq9NEB#R?_+{x9;;iN zoXXeVL6NS*%HSG200LCgcpV>Ih%CZ?{P+PB<1VP8AxIVy5~AX>y9xES;B<6;RN1v? zSwm(D@cj$OG(n$$9G6bJa_!nRT-R+dUY^u; zP|Z=#K{q@wVE@mu;V9e!fCGi%GL~vDUc7xuMP}a&h5QghOf`(F)6ZUUqC!7J1T{z+ z9KUTMY%e(I8XdVwY!p86>!!~I_jbFHn-TR`D+BC41;&<=rjO2}T03P;8d$ViSP1Y2 zb?$^dY+BSz@6ytLe};E$gh|$7*u%z$!{Fzj9iZ6F>F)v;3h%W{yxo_j(;u{qfcr)v@b~tP{ykqy8KBnKKH{w$3CRwC? ztL-TtGFjD**YfR?`B+!OV232fAA;JKZv$vQyrd(&4DM45A zG*Mo>^M#Jv8C`d0|0wxb{#*TJNIcEuiJcx2CHs1h%1SoLJ=3q=$?zc3nP8nnLLT_wN?_q@p($yuW_c{mv)ZB6}cigzK5`FxvXhgaWGO8Vk7LKYk$!I3jWU%O z{2e=&b^q~17ne@l!N;Y#kTL&1gEG%m^37>U9 zzQJ=3GO4CN8&A%jWEttqPDaWkmz0G0Qc_Xn#Y>u|K4&1lf=$57!}BAzNC=9fl$7Yg zrPo$-zs8y8r~-=D)RaL&2&(th+SG>r-aT1JBb^=Z)w;&+Vj~*V15bj6dWhGmLSAzM zI1$&tQ^&5<;CP++y63s+f_#Y5eKKLcmY4iN$7(7ZdSPs<@-iDy&+ge;S-o~MEh|g@ zA%Xl95=CFyz-6XGQOseWWu_)!9QC;UEkvtn-lJU9kU`oU)2US2LR zWvP{4)#oUp2wF8v7z6|G1^e|gEcD<@%YoBV48O;leD+AsI#I|SgS&j&F^mozBCoC- zLO@(FZzO}IeRAoalZYryI6cb7^q0cM$oii#PeX;9X~bXp6muKS?3F|C+hbt?0CLb@ zTdCR8V4GtOufev8b6U2ot|`nK71uFa3Ui3u0T;ksGURA%oTZReGnA?JfY6U)jNQZz z;&8~k0S{%S5%JHph&)PDCA$PJpudUzkjQ@US3rrRrlvyWH6yYA?(Qy>BjHo>eDsAH zse8|&Hi0+cYWw`Jn_n*lGEJ`pgoGptINKN)%mF7(Kt!|ye`N#*8`}>_8)0yEcGe2? z&-R}-e4#_&7fm3&4apF;r@jC2{GcrCQ~E?Vg%>Y)oi?ZE+9SD%iy?lMpmt6zD%ze6 z5$4*m`HyQq!?i${`@ROGcdc>pQeUp7&-h!^(Ma#VZv>>5F3RaJ}=?>`TK;p{(G2Y=E6Spq%_h*uvy@cJ0H z9;5TGF0e&F7vL{u_+@_iGQHJGQ*ZVA$(dIGDJv$#`SQif7gE}WGjspCTAlu@UkDAn zFNS)9TFBwyVMtdA07VG=FYe{bx07L{qoSgqj_5TljqB>y%S890Vp#vVTO>!{1WOK? z2;4hFi%`XR>(Y0Kd+@0R)aXU%2xa2=>;Vj|Nz{lYzrPxL zM}p}mrv#7Sq+%&0Eir*l8qdrHm!eUBe!pJpfG%`2ox-~2sLYlg2$|X{pUIr(iVD@U zSHEJro$1tDY<3bX&e=jgu7uvLh3Yk11zLKIe3A*#;#?I+*bP_8T3gWsYmCYV zRotKRdJdl4mZ>E}|7O?_aScngp9|}8b(Ku2aYQDi=8)N~k;V*1iranFGK5FZ_eGr( z5)jbXX@o%n8q1*nZ$?sFg8;EkaA6?A@~Fv)*@*Z?HuIF^!ULVt^SI^@*fsag#1*oZ z6K9Alvg@W8oOOj-eq%jY-3Tl>Ltl`7UsaxT8M(df5P*^#-!7@WFuPfrIZ$Fps+?0A zM|fe>BUpek)97P!AIuS3c?PtLsA z(~|w=GZt?y^xJb_b=1!(X7r4Wjm6%@N3dOYT&h_|2ywn(c%Y*(vCDNgjCiNb&qcl3 zjv8CQ>+anMlBez8?OHXSep1wKPuMkW)6gSh$gmy~DV=Y-$G9hC|&PK;1=Z7DydW@W&TETW1YO09#=LK9O(iah5 zvcMIe?lC=aAlagh>(oOg_+IEC=CB6d7fDP^tfbiS!N-T|u?ymPNvX_z9KB>`?-yHdY#QGvn_h5xIF3Pc5RTN7g z%W!|IOkcWlD%E3m!~Lt`&-{ehV8^{BB-rJud`wAHCYpY3^@$ z6nJ7fzrL<@k_b6+C_Q#Y9W6joZe?Rb*x+Ya5CwNjD|gMe8;GbtYllzIqMqTJVRibH zlH$AeZw`X%EL`I=slGab6A zja*N9v-MhbXeos$N&(5;z z|M&<>pga#XNMY7E?*MadWNAr5PR?gOS^^n|p}2S;rqv$YOP-n0^UIVM8@g;CZ!5lC8YzNefOfS&aO53_*`~;l$EZ{|T+S;~J#3ZCcLWdNP4rv7hX=wog=>};L5EX@?MY^S>yIT;XyGvTSyS{tQd7k%s z-}&R5pE}HJ_F8+b`@U)+=Ub&{@JM@!+I=ez5nhBFhF?#nc-s7f5)60jEn9LI+bfNf zGE~T0Mn^_o3Iq=!=N_jjNr~gZh8p}*fWjWHU~O()TOmw`g~|5rt-k2Fu)@8|z3#aF zoDf3$tQ^?TkcJ#lCHM~zD}>@8KU%u0Yw=XTD^dRaUYDs@I<3$X6~-{>R0JYQmd@Hm zVe{z=RaII%pA@Bxix+IXsYx8_lbBIX*c7$Pilb3n)Q=YMz;ykvkXV#+6_* zTo}5}K!CzGBqb#i{^8F!_fMZ8e>XN7Z2f77HamdfGa_mJY?)&K9(F}VMNX5hVkbl= z)Iq8+)N!MvuSr(~NEgcRN#q|n*_W;MM9ChLC93RXw2Io|`aH`H5}8|&!}(J+xB~$T zq0+QOx&^)~E9RYQX{Dv5k4kU~$}7!vr8PBeQ^lM1_M9kJsOptC-x-95hvNk&CnuNS zS9sG`ewF3j{Q8aF!{pr$q0zFQ9qB|=d5o}wdX>(dAxBP8F<6DM1VRs-5IKq(8c}KJ zy%6v=f}P#egd%!#VQM}lk5#tPY@hb+C)6{bc6F_=HQ;NGfwIGS+8rv9a>$}HtG#M9 zRqX=w4vacPjB#ed*8tNM1Iz=+S)qRDXl_o+%v=GgR;$*nLYA(|c245?^P9M~$u9XL;(L)4G0!wj2XOm@xslmI&HY)D>)t>9zY!Q z5$&k4k{kY@nR5Yi3Wej=*Vls}hR!`KI#5a*?o;VM_08M-#RQ;AOuAxw2X;aKFofwp zzp!u)+;82*2@|Sh136)2JQv!;!g5Y{n-4OqRW@_V=Q*QMW4;c>ligY75Rv}cnT(ki zI5Gs3vNH`sUFZP8^Nfz30UYYwnq>f&be1`!56Du6KU9X+DpDp9~yn<|WZcE@wSKWb}Mm2EfA*&vphY9n_j`Z|Ld&Ed;slFw5}psAQyj znOMf0l?IL$nY1)DWrH{lv`Vl(P;wal{?p*8t-Yyr@UkfP7~K42mZM6iZ;gy-ag7X3 zr%~g^&&jFGOjKs0>#S)S!J>_1c=)h&i`?AM$Y>J4s{<%NttZDo8U%R>2JucsdZkx& zQkCN1aqWS!ADe>T<3iBmvJa`Jr$;T}{}#CA4#$6M_YXs!SeT18X4A>@EcdjU`?E6o zfxz@iPM#Ckti~rMJ_JPvYy`ym1&=v6c3@unA(8hHsY+215t8CS(hZG?y14Ll@9q=d z@964UbeZb!S612Nc~3<}g^*TI;AOiJ07LsoxAv?#po*2Oy%3g`gM)*Gg$3ic53-vE zu3ms}2Cfh_M**A3H$qCgtNcxko17I7Cf&On7SpPB!>Fq$gm9&LS(`k^OgIo`g z|D{{Q^9buu2)zO&a3sK@!fJ_94M1c9tjjSmTA-1D=UwqI%GK4ih&d)9q1*OV9GNgL zuR8wg*Vy945akDP3ay5P|4O&AVS)m#SX5Gig^g|Ph~WmZ0SH>jA+^j~#;6#NeL0b* zD;QKMS^6Asn;OBs8O#M3s%rEny zk%McFhY-%`O2<{MU-qV^rf`9PNXAW$ae3}?M9HpVtHFPZW?s3uxwQlyu6q;TyC)+K z1Lw801u!UHFNBFepomsg#*NB%^ajKQTGelD-@YwRPhWzS45bc_GWIX^oyk7%%~ za05O=Tt(5=`Bu&E-_H@ispqW-x`D*#;^G1bAX~9@3!cYX0H?;y0nj;TH1%ns8u~1}3IA=I8h3Llkng%2;5LdhkFG(k+W?jOE6ZX(A;9z3L!U{6OXR7w|6z-V?iv!#P&_r0g2jQgkr-&B3$T6nP z)ga@@=mc*}(w&6MGZUC{K%HdPU41IUVL6hIqQ!uH8In(lQeq6d{pBa@GGji)667tg zTHvDgAb2_P=O`K(rAnuYPN%Ax3cDen|C0o3k={>cR%=Kr7PEHCo~ZfKIQhwDX%cH*ent z>5{Cx{Gy960x|oyH<{Zb7>L|+psd0|L!YPi_4M?dot;%z3x*rP3c~NQGao%PF);zl zW+SL`L`At`x}j*WhDb^2R3(;(8`rPX(b3tFFU`yVogN7Dnk>E%R-fdgB;*g!t3l}i zr3x%UsNfKB#At5Ii=I`TX;AJ(){QNdX=j~6&wV8&BVz%T@7uSi4az+4waII2sHn6n>=t0Q7;BAv8Jh{W#G7m<06^x-I*Su_uY}jZ=!Xwza6Vs&^=+Yhz;rMZ@$f zh{%bDeI06iw;=$ePuT4$szKou92_j>h3SJt z*4O{LP@;YC0iX^-a_kq{2s=yl907+qGBK}D zNC~I8_N?EXOZj$2b-RtyZMoxEJ#Tv;Jn-io`rn{8i^Wlz~ z29^u-98C?d4){ORii%FJiCErp%v6~a7#Z@TlYDW{Ia#ZnG0m#iVXmu&T8SOYOhMt` zK~%7A(=|LE!?pLk%i{urms5?Aq7-YNgm1?PG#h;Fjfxr^bQrS@EcD%^uq{kw7h)(h z*!`;wNFgpsGLP{kxW+W~dWB&HfT%K%7-$OM_A`WK+NkYjv}(%x(0*HH&9@ro&Yk&} z+HiKBtLrtr`N{lL?PAB}Q96mlSyO+~dmPd!5)UT%5*MLcj>O6|2z`y#3H6tk^%9Gt zv0K9|S?|vRS8_FE87moaxGO&YaM7X;Vq zlFkR_g#NZp1+*8qDBA8FKdp>PT?ma#t-w)sHn{jkGv8x!mFHn z&m$?y?8KT`a!f?bcsB>Xw@Mfqt9ktDJK#*&7g+f2#AWs3PWFtSuP+pAaoTexg-?=U zvs6`A#}9^v?WN@_-k19NC_OH$J=0aren<76hFZ#lD`CMXEnTVk_2RNZ(sf^6+&t{0 zrBg-2Y}9{awky6*rv^Xx{txGK=6Tao?9?#LoUxODgOxbsTb=XH=;oA>j+?Z!nQHtTA;XzaG>1PJ?{`X^i`6#8Lv_M9_fI&Uo({if^JJ7iUuX9jf*+rkPH@Ja z?0qMlWkHGsZ`*4z^6li}X;?Hpq+K5rz|C+Od*d&Sb>XRzlT;#EtPUTf`9vg+8&N=2 zgw3r|gWqeht((A9p(k#?7^3-Pyqm{(k!j?=T7V-SVN!*3eCB8zv?q7GGD>sAQa=f} zO1HQqCVlerDSgRbV%vhf%HO#=;X^IK1w(uDE+gUDFMF@0Awk>ppb(`D>L>0yB(|}? zcCPfSKkxk#+4q>@3cVjo0!yuNftzLw*06P3SYF?%#7y=r7B7naK1<nghZs;lPSgBX-6H2tdRFz4+ ziq_Mok@O)iH#dHl_0B8omAMe>)ttXapJT$M(s&+-s53AQP#C<`f3>EH=o9+ZqHX!Q z__CpaexeUwQMENa^(BH34gE~@d)@Hbi;wKdUtjw(k)BlTUdVQEr#z&Uyv6rUf7br_ z{)a=VsV-V0bo51TeXRJB@6OVn?oT+|RAXxl9uK$SJGPvun#ReSPbw(iZ(~gRd0WN^ z9jzycv<9h74RtsqYMZ^+sdCD}$i3-!GnMwl0J0^TBjcX>goWdE-z#K``TpZRVzO+M z)b7_}*Vpu(tMPQj1W|)7#y%gLGcv!eOb(U&`1xInM4HKGnE3auijvgE82y?Ei7)04 zPAs3^9-*`MW!#Mte>S<^P)S>*uJ|fBen}olayH81W6T>O{C-^YO66&8xfVHEyfaf+r0Z~3L@kqDS$F@+(Xb$;J=@}}F z=TK}jzjING6e4(I`P{#(;oeWzcgi2*uiS2K>FybdW@>Ji zWMJ;WHwd_tL(8cg@%}5wdYAn~|NHhtTqy0cclx4}m@D;;Uqg}u;`hg6Eq`BJ*$mj` ze64WH!}Bn1%iipbkf4E3qPc!Z;_yE4`1S6}vEDf$|F+zyX!1l0`M2sGE2M^|%2qX| zPv9RpFopW3>#LfO$SIrCPccmyiAC${S9P(Y4NbL-xvM#kAM?H)l9!m<-xCCxN#cw8 zG9t^Wz;8av&ohnUX~t8l3};o^(^DGWzLid^sek-cW;_6O$f0)B?y79r<>pWE_gxINm zhYVwqI%Vp8E1`K7r@fuz`Hk1xu|UKyNc7a$(%%K=6*D8FIY|76| z{PExI_KM*-OZb##%=Y;5ti^GrK2ApalO=1`>;9X@(_4R2ga-!)0VESQw1C(6Ryo^U z+}=$eA6NbMENMaU>PPg9_I8~fxe?ADjl`rG&suv$%Kqvz8WC-a&oBKvuolB9;dBVA zy7A2C(?wwNh%6&BebncxlcrC!Rwze)yI!-NNDSEyjaIl#%Qvn>uS+(R5!}8XkwJKm z)A=&qo>?)tmBW(w^lN&Jx5BoIO6IB#q0FDR{2|ZTTSGO($ra)Q2ROf8ckY=Y?7Ipu zwp6N;fIxPR4F+5>T(>y%*vv%toLwdU*QQ*?y+Mh=76siMv}AJCQ=ov9vuxPWRJt-v z(4kx}mJF*f0MUqVt%lgx43jx%7}bP{AId7>;1q`!{7|%zvtM1mg<8o^_F=_rO(cP8U{29DA}JiXW#h8$O3=4 zs1ITpAdFCPlF-3Cvi~n@P1G!kaR$;(fB^Twm!z@yU%vd2zpe7$lG(=_V5{JFT7&W< zY3Ku_@WL@1Kjk5~1oi59l>=%`2&O~@yn>c1JI>Du)J%fiV0tzNKi4~I*quLN-1(ob zqu3al9Iy;(1KI`6FT6h$NcYXrB1z>MVEwTu_=~~i4Se&b`?9f*dw=~(EiXR;)jA*u z0GI(YxC$f*pWQ!(IW*JM+p4J^^9jC>6qN3~$`9By*fm}9V{(k;s62( zd0bS7gy7=>(qaKn3t)c$?p?>t7+isiv;?0A7>gPH{M(F?c@rOnWr$QK>}g}DYuLZJ z(kQ&jX{UM`LGjB;t)r6wf{+2(PyksDTnm;D${h%Vm$tTc41a+H+d&0M`@pj`5E9FY zUxLh@8%iNKs^uFJd|w&I=`WctHmWwEYlfr*`D3b>47gbj}MaRIRU19e||DLJ`UU;7`xYD zU`wb9A(Yrl58P$$FfO1I3?$<=hXXRAl|fwONAzf=H`&|M(=#Zj`|AU-m29v0kdR%E z%R>)93bNhV`1#*~+5&(!pzOgKt5Kl;OjUJ~cpj9d;Lwr<)Rml^9Gdb|Xn&1OOjg~8 z-chB_RYb;RV7N(0T*oaz1?3{s(jN#5UrceobOK=V(|xG<8=C&3V<{H(zTK7EY5GPz zuZXsQD$UYrg&g37C*ol1B)XP(unib~cvu+fbHME(jsl;+A?OBZ1zh0X&JMt-ATHem zM4|=&a(*oEoC;nX%>l^*op5Jo01`ZRfC80*H4}i@lcS?I1%72pwk2d;gL@DTaYA6rxwx~H+o^=;^Lwx2;d*Qy;}gr0Q1E-*h~THSvDQr zk{tA4k!u*Zs>7TsaH|dgeayy*O2@2&26%?>djTc2`mcaPGjVXaZUy^g#P&Vlfa5~E zNP^KQ;8#S%#K2HYgN_5flo;TA;AZIP9zKK&4;_H%Q1RGcI0VUxp56>7Na6USbh#L{ zs;<1wAOHQGr>3V9F5;f&J&0%rg!K09+u+JI$F3|dzpH3$oaXZ%9K!U=#lzYpf%^&? zst|KrmjQYg5a8WoBR?CJsqxV8}$B zYUJp+5A5ozs)n|s5uL$6tljO6v8l~?_`t^$pb!`p;7uXvp>il2>`9N} zmm-)oIE>l|o9Um?j*I{YMfdgcKx(5Byz*wPT zQ0C>U=ua?{0ir>g!c2gt-4Ppw9CJ%@5!S+H`9Hs+)#Hz>N@2@b?aD;lj&#u)dqCf2 z^4S3p3aAY9WY=%p!0(0sBTeB6aHM=}F~p4Ee}DFN-4kXPfH=ACO$`knkto;KSkd0U z^vfi8zD|y}1U>%LUkQG`2m0UvVPQ~RgQ^Kk-B0@i7b`)OnZ5f)E*`5hiFZYDKS4&B z&5`f~k{!;^cDrxgx&^!f+@JwjLK5Ip0GL$Ee#ve({~L%h7%7lq;{^|?lmEMPJKSZ@ zt{F*_&FP_7MA%xc$O2osxVY#9E9B?TpY!wC5GU}9FHaxu?roZz!X3&k`Wj<=wb~e~ zgbWDujQ$J2{S5T=;VTN@Nd%D#L_8Y;630&ztDvBu&KL$5xuZUb7lgT7q#yqeeKV|6 zC6^2jow$mLzQ-J;>SYBNhO*?J8(3sqR%2WU-=Lt&Kb|;THz1f@&&DPpI+~1#Xed>N zSXWoKQyj@32XYP6>ou}AFwWt|W2BY9Sc%yR*Gfl{dWH%tBH^uu3PSr}9}H^)L>Dmz z%zx!6ieX0`)6uniXg!dp8wKv{IRGR!U}u3197XOl;VZm#aS;MDVzh5z?ZAKvV&~vs zDA?Y~iKlxP0KI)US?cKMZtzg+akFq6$M_g*Tll?jSQL2X1$ zzw&$pR5(VXXJ>AKb1Si+TuZ&a;$c9Qx(}-}>Fu|OXaQ$y2M33BN+e*AaNgM>3@;V9 z!2L1u85Z2}Xj9$@Am?3QtIIsk+0+jw2r`Om$t z-M9fHRD68yi-DTDKfup;fE(7&&(DY>Dj^|3GbguZv_ittk{Q=*1LH@sZIHQn+7!BO zjJTbyhr(ObtDJP)J6u+yGft#^+Mn?D9jPV`9pNPFnvpK;)|EJ6!EW7}-`Z7(`Gv3d zK@@Dcl$#Vkqhf>p%#*DBx$@<((tALN^ql~^8z z0wB0v{jd>0Hd!z}9R!*Cu~_M)Ys;XH&rKuSRt|6zR|5TqbR0p_|gkYe&*ZVJenr%ylD zoKMy%{Yy}RYoNblEmw>#1R!5l*umiuzC*b3!pCjD?N2YC8Et&%S5+pxH-S9 zF<-Pj?0S}gFb73n! z`QNQd>qj5?4$2%ZECN<>p=y-5Yifc?QR%c>$3cb;@ z!${Bsk(Mt<%i^Y9kvv{rilrWIv&q6>X==8CnLLvlPRR>`pP#TpqHK~02_hz4SZ>mq8o+y7azy>#e0>cPP+k-G^ScnpnHCS2mU=@Ya$-s*04#h)s+&C<$ zFicDcG6+o=lFWHofB0j??`9Nh%(5d5%n!9xRO0T7bldJuWL^%_jv0s>W_ zZl&SpAA6<#FlThu^zBW?spE0R$Rmh zn!h-0u`a8mO^ioI8%zDlWOcqfV)e4 zSj}oe;eudp^WIdvTknpBS;w>nLZxC*r0Z^K`ZYHv&C~d1G&FvP5*&AkE83hC85^Gq zQC-(9bps$0B&Yd)e(>b;Op18KBVTE2*PU(%NJ_Q~hVGJ{e0(@WH zAx}&>D^H`Sxxc>%>K2%(KxhT}_y7|DKIGaVAR9U6p|nwBAb?t^7(N?HaL{1|Qi)7K zc>~|~0>2;XU8t*+n2A`y3)b6P07_Yqe*^ipy|r}&#W$A_#yn=5WLZN}Os*$k02A?cQfAMrHKRdlFFzGUttA+TDzXxW|n1OXec< zUtcKwq!4~>swJIzGC$4Vbw?xWtb|E|?k3uIUHOL;O09NkYU&m_p)q$YKUIa+WbSMeu)~yT?Heq1z+ap+EpGt-s z1tkiQuW+wGT^uLqQaBZkl0}0a8~De{sw#RK8c-*Jyc}d^P%NuRHk`q)N(pN_IMW?$ zZDG>M1F;#-s{4Lpf=^3pJUSBe7ocOJ1F;be%{+*xz$Fw3b=hE+Iy7dP)CtW(naga^ z?(6pgmX5)n1TSRr`?2Q@r9{`9JDy*rB|-Xz}@*KoeDZVV6wRxnV7P2bLkKu5HASkeE`4w z|Ck9ZqKeRa&Z6FpufyWPHFyfVDbVnS1_sKfYha|YfwUOll%ZOW2Z|a(4uCkF&m5Uu z8apVEeEJg*zD|vNE@Ujg!EopprLuK_In9Bu)G-^OgN)D zdq$Fe;4nj~&5OvWz6|!0)jmweCBn3$l^A}s@u|2x@6hv-3W}_7ee)}qfKKlS))};5 zQ?Qe_tWVX%ASqqf*Z~-L6a+HfQkgQZE)-qpmcz*dE;4#Do}|R0q+0ZmjT%fFxt%j| z^Q|gsYO-B|QG;vO(K-~ElSaj)SbEv?fBrLykYJJsYVuH5&LY<4PPudI@*N4je)!81w;zfgfbC-_b;=}C_ zd>=hb8{hq%;F=S)CiT*PN3x# zEX=eBRh_HSN#(h}_{Hd3eQ>^~qJG2HQ|ktANyG_;AX)IQB2~vg&Ye%GbE}MX_Wl&E@QYQp*B?c$2`MtsRPj#;VVwb*>S{jYipbZoa-{bedHC@9+IlNUbJ>~ zQB3S_io!g7;SFsmz+yfi=S83tDi8}$;sxqi5PB)CK73k#kPbP~N)PW1eca0?H^D;S zj?;Y}`y83dKGg0_c`wUD>j4^?{jLLz0uytSwIW{dQEDLi4-Kb)z}vfIgP51iz-goZ z8n{X+(D)x!!pjf;LTCp6%Sr^#3Up0V%r5_Q2{37}FPlegZ7J_D}Dwt*-;d z*A4CS5Kewk5h#DO#l^36@1hb#{|f+s{{`N!5Xb*a92#!SJ9jJ?y!0YX65PX)xjHrG z|Am0iH0j}<0b+_+8O}@C$OoJ4T}4sRPyhF0xW|7%oXH9{EOXQdFTMW(OF^PxekCu8 z&jVaalc-QP>ZI@A@BbGy{MWA}0-dXPF*M*Pse581U|t{`3PyrB-R%Ef1PVTGY4JwJ z1N4QGJX2FAA7}*5mD)@T8~(lvFzxDUm1!^H|9-l_JWfqp&$XDF=nIXRP`t{f%BjLGj%knmG8Votn{aMcgFB8bwp7U`T6`XzIy>8d7vGl zKX_amn=~PD|Cjj)FD8}H#%2>N>ZsfrIM=!y3O7*x*xT7bO^#9kL7p8HJJE#~XMo6o zmOdspSW!s{I&Ktd0(melZ{L#ai~yvJvb)0<0ZtEvAK)P&baHl1PSy?_|Cym`dhu6v zsJXZtAQQ{j@+-0y#DV}I;Rk~-IV;NovLqoZ7U05&M_NM__9l}-on2k;A)KsyUQ!th2J+tvpI_9%yRZ+6o00{Yi9jm z>%gHK{fNk8F7xX*=_bxAA1#+Q=xQz)X%ph57Zzy3*e|a5*LXhk0jtMs(>1R-5?<%q z77B>qhdCwJjm&?I_KhG^qmoBhCJK_XntI1bCsi}m&LZ8}bb_s0+FD$&Hr9I@LWc@7 zcUeQHGUu~uQgdhITMC>l#vAO(Br&14^`%lb&-bjv44;5ro`YJB|(R$w8L|Ug)Jr&|1 zo+r|1J;ZSRnqx*qqVX$KRrJ#nO6>VfX=bOpL+RDg-*$gUKXc-^y?;L{Wm`_ee;aU! ze{%tDi97o;RDDMN?hiiGzQro?q}`zF>eL!2UMozxDJx!~f;$dYhl7v=w|@b?4>cJa zNR|LD3!RkRyd8)i0QCe&mpDBx+~lrYLXSsBLj&dM41^gK4ur4~CXnP^%sJ1<8+YP1 zw@PuXDJs@^`(IRMs$~U{&&yLl@fvMlS%p2@l3-X}-5K`gP9w#cnER}alKmmJkGSRZ z&DL@siLz9cOq}((u2@9a&IYcRo`4f*H^13Ba+b1#f~V$?SD3anEzLzhp!bjWGf(fR zfw~?$!mM^;+U$#qu3B@QR6bA~-O|DF7_+X;>zMTQd3coB!cuFO^!MA=X7FJMZz@qnQd1|_W$_K$uzMZ zb97oPJ?~=F+Vkg$YjTxhZd!jgd#DBtZKalZ`DlN+&fbg2UZ1cZnZ#5TicaaCk#(nk zlufRs?gn6>Yd3B;jC^3l?j3aK48lWHMn4saOO5}y*Y_ql2E6@en+IcZQLb4>RaE}W zje&>$XW5fiJt}&&*X!t}E*4^cOW?}RKNBbRc~3xiqfAbR=X=%f-#8NS>F(-(Uofkx zObd)Oc+>4CKS>tJkICC%1mRE@4PMK%5JEpg_om#MB(`0$mS1NI#S%mE))cO~0NJ?T z^%72n(nksfpWgF}5i21t4{<)c6i06~6@5HPG96&Y9`@d+LaWB&Eq@<=O@=Pgr4ZwT zanh~2zta`=^wiX1cdwl%VjY3P`r1M<9+(;oIS&fkkF0|WkG*SX{-t+ehb`s~X|w-9&)@3wI7)vGaSY4jYf`D2k* znWH}K+I!an-I^{z;=G=Hy}lIR!G#O@X7W7M8ViEVwr1VqY5EKivcWNtU+O2Zaqsj0 zSjs_T4ok|;{HC{dq*jA!H1$Grd1%D$&?b4zbO7PTeq|vIu3c5#w#8t8=$!D^OW#yg zZ-%13wVoIYd+WDOw*75=7StGxFJo1sdX+oIA=&b|pMZNwx2>+UuWe@ucHa7rjHylB zYn74Geokb=D)EJnXD2*{Q!X`?|Dm=-mdKYK4=T!i36eFTIV!1X@w6gnD8Xrayd+Te zu4@FpecMe8CtIPjcgg$jq2HnErnm#A-_*3su>Wbjb?t2?(@|^wus3-yc!BBtWoDx+ zj10WFCCRpXqaWc0chd4Cq&!OR_-fT2qADxntSGxv1H&;uPEjSbxt^#*Es}p(E~`sb z3m=d)p`_^%`d+}s%kqjpH-JH2dO&lU{u=6t5!>xOx(WunZEvdg5YP4`ZVt3qsAxd| zDH9TOpFDZ;;zbY8ouSgHLlO@Cp?|^jk>mb?G3?~{7`GSTeFwnE}I%IX^*WbvkkqnX=kGEH}VqZ5!xN^SRn>|CQYKWQW23_(i_SWag3Bg48%w)r(k_xh@ zQU*Oo10^m}g`aJ8jPvH}$>DY-OQ#-$W>vV-NQJ5dz zH8Xv$DvqHY5nHSOR^OFH*%EzIkXv#!NJ+!*sAl9%cEe@o&)b?F*$up!?#o|_A5}IS z8Sp+NYrJ+ihoR}3(!eNG^_l0=K-{9NaJlRbR{Z&#Z@4_~+n$GfmC^Dya(io;2ts5R z9?go~!W@XaTAQE$W}@5{sy?tguK+e=o}iR9m@E;XS!gI_Yg@1g2YR2({_?xrO}GT7 zlmY7=Xu=^)uNbP25*$17h?p1|@yg)>2xXy3f^fD$zFa;fle&{dAAxP&WpegEfn=}J zRaNbJ=^bN<+uF-XsJ(MF`8RrmC$FsZSL*bQf&5~lFc|KNF67`LD99-Z4e+wf(fGwdKEh4R3MkWTbwmwtXxjhq(tg|AB$?$4qg_ zC-|f9nWoZ|hB4(`Tt+rhXeh3ftk-3lZ1f6ywih4GiVBxy1)^1qov*g@Vp-)nKF-O* z6riv%RkkI3ShXwv_a^d@!dQsO-;Ez4;SN4}(%X_5O!qCNR!iLh$XwhdZo@QMp*rwOKtADI(;x}~AcGyDn+p;?F@=073GFO^QMpCz z$<=jq;)8<=fCvWq5U9m6i}cm_Owj8iGVA_SFCrH$DIok4iuW9k*d>=#B@G?xc|(p6 zDeL9q5~ABXjD7vrh}CQ43bt(#?u2oSETx4e<+#V^e$ZbPo~`y5ztvHi`cwW_yUqNW z1lgu#{i?@#=JV-u?6!Nu6$f#cQgUv)Uh%(h02Z3HxqaoSSMd<`A=J;x<#Z;rliF^rdi&O z0b>3A5f@1Uw`=%SdlZFoO)v#|8C`(v>DYBZg$+Gl%8xOo%z$vmk z6w%N8AhGdFllc<%$)yQj{OrbG7lWy;)St7rGs+jLFSzS2p)=#Q`*No<5@FL}#ogIv zL;N?W-4Q4AG5?!@zT2iV$u!opsb3+>pad9UprD`)OFKa68epFVA{gA604M?taVFw= zNHHQld3oi`K4LUQs{a{1MW{SO(G!hN7}p303HFhyiRZoEhBw>sJ6kv0brz4Gj$k=J%@GtRdeFm1hfJE2w;d zW(OXC7qT=`cXo=Y$XkFu&<7#{GK?W)HxP-0D5prcLGT!89{l!Qw7V(-6te5Vdq{bl z0@fJxFpgE`78b;hG{wcixQ=S6P!`LuhzNLQP$Vv@4}#t;rsnfF+-3MylgrDQMsMh} zC%fAJ0Y)YI-}u_(sL6=GtMN>SrGQrCY)foz{o}`vYFUG(dzRA#iPyf->mxg8fAQe< zKW>+nxkf*;z_h2nODW{;rb(i#MrT~hx_i85O>JB&PQ1z%_I>By>5=}2X)Y-@Y*3LBFvmx+rdd-GO#`PQ#3@5M08I7Xy z!yvqH6vQlmDyPUzyhCMx!>JFYFPUM8K->UJ-z4Z#AlfYfOt%AARqtUo!sM_*+{VFC zkd*~8Ru>LwfvyO3It}2?x@RBF`ErGlAWO*8-SW7oTeu2jGO$LsG&c{emy`wI_W}T@ z-wkHi^nUQys)NBdDM=rMD4p0`Aeex|4Iz0J_QrDv;^Va(xxc6yFq}}PkU3nPl_Oce z+|4g@a`E2aCvWqC{?Abc)+E{S1_q4u=>;DoF^dGv#)z-TMKuyfTQ;hc`U7Dt189Mry{7d0nfweK3;m$XZd+zn+%fiwU2y==q4#1^%w--inW4;8k zo#mm|0>-s#ZlEZ5S5y1_`)j~u_~SOlZN4mRyF*4Uj6Hah zPzPC{!s0;|;u{z^RcO?{*co_ALn-(c0gVXP)G5gJaB&;iK(LZk zC&a>H4cH36B~xX2K9Uj$!LzMK$4f|>`3Wk6$22r4?@v9@6B83bF$G`@Ksftl>Z+=^ zij%uvPT<-poDzW;N>j?PbejHXT=Y!*+iq(sEuNg{joWuOGJn%M{6Surl{(83QS?_I zn)1Y8UH4oHQfTj6wq#y)z`p85(wAcN_0h-W+O^i(xVk{idc?`}vbopYE z=0*v>RsR~xK1fWfwQk7h6z~T1=OZG&8jGMTujo@Zjh(Dr4qV^Y!5@vx1vavAN%G?e z;}h@f?D^nQysNDr9dUz`y!DiVZ8%>w>So%7|6)oBCCR}J@pKIAvnGh%me0!QUH;Z| zZHUOB_pVQ5a;}VINu&w&q?{Se0AE@R%d>f(hKJoR0*VIbTxPG+6%U(zu~=o$UsxNt zJG5H&P9nK?N-#?)V;=F|hwQm&i#1X7J48X;tNC&Nft}-puD}>sQHmsV!W<@VIP){K z)$j_!x{=He$K>#qTuUg)gW=`RR2!>E2S!F`{{B$Sr?$bGrG$@e+pv+WS#6R9G#_X7 z1&n$WYP*L8{U(f#O8WWa3n9Vaw_*AW6@Ex`2V&C!!tdazF16u`>4a1`eUe28WwhNH zgBPqbPC)uux5MM7Ada%LgCAsF3pP&F(H3yZ&`k+_E;7^#U)k)Gqvi^tIS-(!VFm#f zZvN^5VrJ(aZ-AG%8bVLkAhI$<w`HN|%%8;a8e3s-quaLZ12+!zVFKUW&$W7(Ss=wx z%TRGNk%)sxJ0K=o+#mg+kgCL8(t*}qD7W8a$|rls zoYj9VUas>i5C1?QElspo}E_TRmnIp{v=84A5)pAAgRaSh)e!8Q3<5)@G?c6r=x?Msrvx2+Ra(pq)zA$1f1 zcd^S7TYDj9BiCk$a+3FCm_T<2-|Fc**MWTmQyuraqa_Rxfrnr3-sQfZ67$(?roc8p zwhQUHRQ|=0Ypnl5wzoObdNr0|$I93_MOB{OZH-dl%533hpNFk=>kP^-Z>^?gBi!TALD%bMI;Kq&PpdN~bIcv`#oo6K8yW=lD8xR-)h7nc!ts zwibCtSk8J@6Ps+&CeA5Mufl0;=Na++y{TWHd)qYARRrgU=oCX(nd!W3| zU-UtZ&f7P?r%5E<%{?|hQTk$d8qi>1^v1<{A z`+E9w;%aouLm_ge@8oTK#Q9H*3qF=$HE?uhNpZ#!L{f0XdZu<2p?L&^`H{Xy>idDE#1XF)2 zj3)OLbH3Xqr3q3al%!MKD;EXID=$3n2(S(HZfa}s^f$1c5mHjpz4#uemihkC@fCUu zN=uOkV(V)X`L*<}&bg}b7Up>Y?32OkqMXZ*M&(&9LviJV@S4;YlRv3&WmXkkO|DeR z*R-^1Ojcllc93Z2*|4)=Wqy~T@8Fag6aQX@A@CVzheBufxPIVCAR#36f}GH}$OSfA z%tjD`h`Z6~P=~c{R3`WY((02rxSL$%E)+0k%k5BVPILSwUmj+NP1Z27*H!osi;DXS z#*Oxk$fww@FAK6A`c%{$+T_H)ZPcVB+c>?vH~vV5-|5?(kL>;qV-BTak~gYCrAKp| z3%B)cT*d>v=jgiRD!m;;Yp1AOcULLmjBp>VWHgO$2GxwtjD)jAS$k03@O&40RJnVH z_lrQJp-yXCM2d*L!BM@jgv;O*MV~fh&9Gu-N7mdyHT&-`S~98rl=q(YsxO`%G!kbg zjW#|U2+Os?ab6{ls12gN5qCek;f6>xWf1p0>42nGIR|$Gl1kbbeqT-5KQ@QI0^^3a zg}uAOHhe0!<3hdf+1U6hA>3VxOL^*_roZ9T3JY;Jei?g5eXW=wnmv*rke}5=yuafC zf`9bX^Im1&Zx-ySJ7QddArb;|oA+KkU+%>vST&_rgL;@XSf3sAey zZ~cO7h!aSB;SZXe<2(0gx}m0_PjU(jPxSHkC%_cTAvzBY{w;j3GSd8VKhFb*xQC!u zgCh@7hm=8un%BSVmbm0bF?)9f%^>(Mq+sl z0jcgOOZ{(F?Nk%31}chA?fCG=b^3~MVx~r==G?TJ)6$sYV@!@WozJ%Cs#~8oao)xV z`ATz7AcK^;gBJaIQO0s znWi$MVaFuOf<2FW5NW^ZBKk5s$#HLs?Y8~-!QyC>S#skI)bFKmUMp&Emah^+F<>UK zHe&IqszOt>kmo2%YV7cka>Q9e-pdJ-wV-t$7uN{qaHZ$z>W)#VI)b(&d0+DH8_F=% z5LxN%#DP5OeOXf>kx0zm&pHE4!94SQt%ytxrSG5kC0@w$)AE~U;NB&A&N?9;qtMgE z8LED9NL*w=t{o#`&e{>beB-58+Clmn&Dh&ed1B=NpW-W_cu)CA2Xx++eD;E??t6K% zOR~-{^Ng?w|CwoqEKd3Ufi;TM>e}Nwg`3yfTc@*I6KbDX5Nf4O1gF4d*+8k(TCLl%&IBb#t z2viXe-uf`CIUX&B2;SRZTAmz!6vqA_u}{%pQb|=SQK{9srd>OXeFnFW9e+ZZxV=!@ zH9xLWB|HKjvN&$u<$EA{Z%8$ynr4pFMsns!-S-X(cVT-ORZRBepQk~$H#ft_CRWq1 zDXc5c2zbB#I2MaC#H-|qJ8{e(S)x@1l*x74XP6$5*}d=BV2V7OyOl@$1*5m!z~>@L zEzKq5@lY$KtY+**sMggG85#vc$HexU5>F&vw6@m&Sh$H)b*5_sUd!G#-u04oIjQ`?we^y(xhFlxbBzy zWuQ-%Q@L{X;c?MLd+Etlatpm{@^KwDO)8&m+yR-XV(5_tba5Q&hcS0#YmsI9;#H## zrbv4z3`EsDEt8U#)>l6Sr0AMw(l@~-Xp92?p;Ksu`DM#@ro0EuALlaM)5;B>xAeD` zOo&eZwuC^P0K9$HEp)4Gvs|^Rn=baCNMGK&GCM?IB@_Id2ARhL2YNP3i4IUo(CCVa z;sGrX2vpiO0iNc`*dn5$btfw^umNX}{tX#FNx;e{Qgb9Ha?~*JA1zl)jLyEMrp%a> zdruyI+i!k8@aWRG{O#|$pS*!*Tke-#HtPJF z+-tr#bLfD|zfkz6B(IwiGT7Ag(+FjQN9LW=f+Oaj{OBRfUz+&1EPLd90&oj@>7*nj zHv!XnlWQm;0)$agUTYvv3Ed$F2b6uIPp^OyFJQGGlBH~ta6BELqT!g2OTUh zfLh#Wf-XK#Jq=$y0jGemnxl;Lbl8nsf;ik@#-(8QHk9)fMyuvPj%&KQO5l6vhSpgu z@D0JDiI=JxbXtK12%LVP2Y@!pMiU0SF(JV_N3o<_F&ofD0O~&mZ{DncK#R{$4|Iyc zUeUj{V4x8hdRtWKY>*?xsK*Faz`U{eE)J9m-LploQfN&GQ?Z)j~7q8-lj(@$oUVAK9%9w9MGV`^38* zs#8^iG}2XHuHx#I*%CU|z8u^#|h2n~9lNS~QdL<;``?k7?qi9PzW&n}2a+t_Si zEwnT@KhSo1_k9T|@z2b>0nlivdnPE6fl@=@+4q0??GRgE+Vu~se)0peA1dr`M)!bb z$^r&R)$93-?`HJbk+wgu(NoL{^Z7kJ(GuFZ3}#lc5s}Wo?}xuTg8&UWIy%%*9o7a6 zKn4+A1(QH42AUHK3vT|4s`8Q> zx__V?4fneK;>I!{v`CTXR(E=So(L!oc8K+r=-*(&~ zDR$2=d9p7O*s8hGB+*Z6kON3d$el%4&pt?SRrQ}`NMsC+RyZb?Y z#}rA-LrDrg&eBoAhuoh!K1`98I9LfBz22eh#2TT@=4;t;5x4u@9Xu-QOGayG(4Dyv zqJC*SlRevMufx9n=$1+Z*G7I+|MVk5Fd=5x{;gd6!-8#WGIMtRE9 zpR=WJPq=)*(MlaK36Y?ug$R(TB)zwg+%%ja+<9rNd6FaCFkIK7WZ?bvavJ!^Zx2Xz7RRWveYk%PkFGPr(fMl zC`2H1TDJwZHi%ucLGLsSf}C8!=Emdv+mX&ccDF2kyfiA*9>^)M{Vfvm{Ht1Czfo>} zJ`Pq>P&N@{!Vm_i6zK^7pqiX76@B#R(VG%p9-hkaN4X5s_GjNBB_Awfm&*l^oJh2C zV5+ezVp+Iu|Kv~5`cw4HK8|y{v!{2Kz)hyabDse%Jp==N#gs+3@1=7!-@LH4u9O;G zrSjx%GWHQ!w61YpRzvW;H*e0YuoMlYh>1?KhON_UaC*=3Eww#2@OHmgaOiemv|BS2 zDn%OgUt`QMuB_7}lc5wru_nr1@J+DUUF7a@YueE|-(d8$el_vhK`aQqd8Bd`IuAwoQ&98}W^6SMZEENxYuUZ&*&5_VS+o z#?5}jJtf2wV7@MuYq3u`ynX=g&UxlNi@cSR0Am6A?_- z*m7xG1al`hi$~vXB?ld|lXIb=A|68iG2C-M0b<3b&TD ziOpWu601X5k>!zJ7-y@03QQe1Qugnc#Tp;Ss|S=gP<7<-^4lIg^IB2dDa&J+=31WH z!!8D*hp(>VtCrl?TU7*;%;>sCa3+P|D%1UXI(sz=LS2<7ymeNcFcw^b7B72?<=zimq#CcIHJ8=Et}#m0cU>$FmaP-s*~EjXUjPQt%ih zZ_fzg;`uS-YqY4KQ9Znff`T}{MYC<@blp;jol$5d}eoyuq|R14At*KlSa-cndO9s@92{wh5rzC(t{R0ZhwC47MMug@=hRkD-kEog#}QWPdG4u5 zdExvx@@j~Qaidk7UD>^o5xe3|*6XTNC}#T-^*-9ARvaTWd2Qd4j&s5;Go5Iw=qzD= z$+jvT8fJIeTd4~FydKw)h)sI-NM44QD-UJ~FE)!D8^vQ;K#au@%~7q#9R$MAF(;YgT{ROx&c0k5%bap3vI|HQ7hRX64QCu!Jn>jW3kcY4>u0-rcwo8a5$a+N%Gw7Iink=!}Lyc#%U(uA`CC20V-F#PWf;rCs(xN*%Tb|d4{#z9eyjiPnnHVopo)L8Uc4J;x7@1 zJltEyTpnv*9#sCsj2Y?@4MItLckP-(=tETh*`=%g+Dl@V;p`Ew3b5Di6?4rt-?Yi= zLy=IU^XWRx$6jNol#D(uo4qQ(;MYN>1QHC&C2!_qBul@#OsR`a+NLbJlYn}Fo%5A!~%mVMyF zuLKVTQ5P&wQq#JVm-+Ek!_WQLW7(~w;>5H#-LQv>)5;H#Rs>muKwq~R&H@zleS05 zdxL0JCiN=sXQ;ZCQnR$ZmKail=vWR4C7|dzNr^)Xc@)07GH5u z4{BPWlp1JcIVp`KpXs%AsH>lEM0-e(I=S)W$CVt;x0+vHZ^#-=z9x57|MUK4RDX;! zGfpr@uh%@$ie2rNzijWLpL7MG*Xn}>Y>ZclVkDEuF(cQ@4|S}ahn{0Jjks_3OT}EM z6|_8i=2S&6mZ#9dFYZ_PQHZ3`h@Gt~QHU;YYjRu!1&P>b zdJO@Q0lMAH%-JAHb^ZZ&@eTfaREFXSz3_d(yRvtRQjQFf!0((=4Q=J*p-{pBPRF+D z%Q`d?^kli&W7^{KWIa?{Aad+a_V&KaBx=IpGjhcVs2EKpOI zE2?ryaqAUbOnYqn7+1yw*DQN(+msGs>#X*>(g6+plNZe=fAdk@|Y_q9h4BGrBoeAH#hJJKC{E@q# zyk~UaZF%EI66z+DPsu0noC%43Pkk+!WW-rxR)NJEfSWQkG0p}SDk_lNq@l_?G)cgY z^z43a6(J0#t`Efgam~0*-=~50B;wT8U z>yhiUB1@Gy<(pqYI(emn#9Ec>`DmSYDm5N-cMhvS_$5fY9&?Mgi`k>-#w0&|lXhMk zBkzHwE_+PAtWn>SM)76_YV!b+%?iN^Pq(4dO~`YDvik%{8DS& zaAn<@keg!o&Pp0#ogOE-k~y!IX4Ex1ad(<0D+KKRG#QB=A4M13xW4(z#Ej@*f0~Og z#QmK=*LyMj0lPR~1FalCy3<*_(|5w*q%0QFzNr#*3=gM>vJFt7h~-1GkDA@7Cv8Hn)`~1B-kF+Qy*TKpvheR-wj(XT z+SrrrXo|bS@A#0-=+htTQ*(au`s3cnSbBL0YQ90AF4geXEnzt(dxiLm-3WT}>8VgB zT1RFYgb_yG&uR!X7P(wb7R={`>(-EM3iN;T-ye|VQ~jx7mavjXQocG5MZCK500U>e z%i5wYjWtEHS<|%Mi%KdeT$SA4A1T;hwHig;QwOv;N?6Q-LHLJmeGwg12RET^$$Gs7 z)=0JTw` zsuNr{8ONMPSD&Ic+Jp7gVWqR^ETI~5S-k%FUxVBm&bwtCx;3hBH?~J&4h2v4eS>~M z_FP(S>%6t$kjFZ?oY$CmC-8P(C(~MA3se2lu$V@X|BSA|mB=oGqviVkyLfq!V6~4QBpx@G z$0cKnlWi@$J2kOsniLVX+t`dA=JIUP7U(Z+VvW?yA>TC1op|og^`&LSajSl>)Tsf> zBd<5jEQ4&T*s4??Kin_CKUY zg0Bvzox}HI$Tzc>rSWmQ7|N6#d1#E{N$_1u?76HC=rpP_ z+9^p8+v0yOrAa@{&O44UlYJmPMenUMWa3=#(hgi?<7eE^vsj2Vj3E|2chG{?&OnVuHYH&#=ZScH!71|`zh13GH;PqQQ_Zs$2A6z zY=0(?O~h6d?KSrb?}d+5yMzaGG&s9OhoTxpbchnUZNCt(e8fdwm+Y?edL8Fc^vqfC z@L5c)*4TMz@3Ua&BSdcLw*wcwzH zlsmdbpdK@al2z8%E|r{Joq1WaS{tgcw-9uBf%V;h%a;`Y)lyf?KI%p|Y3E@SkALPb zm*I%rD=MmO=j^}!s8&yNd}~y>^(Q?l*LJGqn`Hf`MUucN(K12kSd=(C}sQAQf-KS6^-VzzSwW#hbjyPLuqx7 zqlGRK8U%*iP?`3rd8oyP@Tl<;luZlA+8%ymlnf-GAJp96p=3Gk`;>&IEuzxG{=50d zy`D~KKjIF*pEnjd*{_81VdU)#F<56Be(=blIhOGqP;j=wEp)j}ymD{LI+v z54hM8Aox$uUc8}0TL=s0SfusI&2ykiNQ4tU_TNp#1fV-L8z(H>*#T< zRmhhh`IhG79{Hvzoy9J%NUFg8366trUZ*$n^SJxQrw;*&rNF1M7Py1UsTp*&!^hm3 zwX=_T{5!dh%WS4O!O(uiiY=r{)Z-B>T75Ymr3jE(r=%iPF(%;)8k%XgR3>h zv>mpc-;Sd+^W&6Mac`xll!A9YU*F);#r^cQMa-Si08o8kf1@6$64%ky1&q^-iYq$` zz#ui{%27VH+C~jr{47Qd&SHvQ-y7%xca=a8@=XcaRZe%yS6V-3r~BVB>z4&{|OLf7Q=7FfuTDco%Qbnf+JeaM(J9(jKP+!e4yLO*x zD_`OrkNJY(9h(JIE%$`_Z2$B{=HStX&B_|sVl!k9hHnhnxn?cybB9tA@ZQmvDtYnX zw!arNvUw>_Fve$=-H85pYL}XMm;G|D#WW-BH_cZeA4TnVxg$~iMq$rstgGa@jED&y zUr~CVcQ*wF8X}SDdL6qxu_oVm_asgiZUk3MfDf3-tS@XM>U0|PTKqtzNnD(JTJTDv z_d@|z9t#7_NRW{OuO_fCapu2CX~0R6latf*J@#PQ>$YB*26UwQJ_k(%&b%GEiPhnu zp{VYaxw(=Y}UKLpi%{}M4jM)^1RjBrDycYilh(1@$dOg;hMCFBZ@UThQWl+CS>5B(5*y&~qJYXU@_vgs3dT;h6TVd|ojbOOS zU(-)ej>&1ZlP$^A9MaxjbBxb$Qd8WfIr~Nzn>_Qn;Y#DTBh1)K9q+wa|6c>RAj||u ze^q(xC^dazQ8KP2)3PB)v6F{#>2+0yxQPwy1u_QT{l{~#a;g3PudSaJ%jkc<_2ap$ z`>xLp_Y*G|yq>a@|85HZ?zmLfHy%$=d479o@f`jR-U~=T*Qrj2B{FB&!~%WbHv9;j zn~)nX#jLk$pU>SucBJ&VK7twd^?w;wkI0jOw3l_)eMZ45bwZVtWe^)@%}Qo zq}55jrgM_OeSYAyrrqmL#Ird+O_*_R~(=EmSt@(-Gav=t7l+f8wJ^g{7UfdOB(e$3mWY zUrq#5&DStoj#F`<|CAG-wkqFEA~iWJ_1{%}8QFCzfgV2zxPFC!T=&j$3#pFq-hp5K zo|3ZF3}#ndPeh;idRFS`ZbYbP=??vYbYe}xy)>p(G`Ug}-1E{Po{QbqXGF#aZeJ&Y zPVH?qksgPA1IS(J)Jj#T|7yZXwi_h;A}(oVDq$#kwY0(#f!!aNId=+HkGupqYPcFg4+DprcMrf6Ja0%k>m}Luhh}9&TsjBFxhF z&Sc!N?t8zv0}t6m^cSn5#b?=VZsUqCFH-Ow$}$_8ltvd_t^yRxy>{fpLsP@b)qbbG zo$@*Z<@S$WI)OWKab83bTRA4Xa=L7{y9E!j=oXFd&01u9ZCz~<##!wT9MsEU%t}r; z!V6xEO;}yi)ojP-h;163g>Vnf{KF^qv6VT`xbH_=Vn(5v4!`~x9J}n|- z(tcFQm74J~nheiiA`sB759<;ByxHd5EWEzd9>(3$w^*eQdv0>@q*lJw9;_F z_?}f#>TM_KU|J`_hJB+WP`*GQbdmMFsn?QAc0HOx;ve4_X5tIp4%)&BHs127|0dz_ zSfUZYRSN%%z_1sxk?Xb6lE2>dUL9;q*-^TR*^EK3?RqQXlRghqBN96un<&0T&)KK$ zv2Wz}td|%LFYJLlJhl9-?SdW~!d%brCSQ6$;(&C5>%DX~BmJ&W>Sdezg(%Jjzt#Sa;-{+M$Ze{7A?$7wm$mAI_sJ5>%+78lpjVizBJQb z$3XkSRK=QMnm=mkR-3b@itXs*$R_0aWbp)+A#)qE=e^x^LTgzvIZ(_3%|y?Nqmu`! z<2{-lCl8<;7Nq09>38S80nNeaiKJ3n!N$(T(vMd&m=EMxbRDH0=bh?%ZU|VJHeXoK zIU8d0)>J4`{NXvzvfpJ=d=a30AURMs`=-8&J#XGYhBj%O`e$_yO%3}F4t8qvGo#F( zD-JbF*PZM+@9UgA07GF4x!!4-J|QICKuGlE@qNU5DJPRb`-2lbx#c4AaB;@mVFka! zx3<%_P(^UZ%$z@@dL$Thy?bQbDt12O-zvrEo9xY%7sS=H?wcu_uf?~raCiD(_5L@Belkm*?b1P zQ*)GX)zGlCv73^CcO^=C{43?1F2wkBB{TfnU#2G_Am&;I)OZ4);Qj4S$mdw#%~j^j zH9V@!QcOg(`BAbHx{4ey~+)&&2VNJYASFX4DqwKU;OoNE9=6)l{wAY*CVuC1x2iTPs-K72EB) zsAe(uADhf2%bSsFpiCzLAnrwk)BV1WGX%+pCi)0OK96niMTvh;xX#Qi(x;0~ilaPR zEmr}+)BcHq&?>1lYI+~X!2)QHh;233R!>vGOyt=$x!MZUNtNhH*U6zvty5iXSlL=u zc6sT9DT)juf;noGCsW#H80nFaNB8<#<>S&pBpX(&AE*=}4;wj4c+8 z&WyjF=~m0H)crspw2C&_>W(In$d&ekl%vu9-m^Lad_?IDWZucbqZA{2&?j|!=9lFb zSP{BKH%v6$8^2@P8gnG9WS(bV>EJq^4@0=EI|-G-28T_4?;2A9hu-;@w5v-wV&M@+DOBFT zyZ;qi}}a8XqcaPEMWP$68wu@FEI3N?=RBdAE20E-}Z8}l`0D^|`I-;)%* z4|G&$OhgZ;UyZ$ye!HclPxYNh=^CQZKlK*Xt$0B~_?haJzdsvL-s!G?Qe3C46w-o0J%2a$(X-wkhrW8!(GvYnQab9nBbyfANMN;041nvc6aI;8U{ z%0Y<|Z+7jxZ0T(Hmb3kO;_EkhkN1C9ws-puL_WXISyWMiDu_@2XyBhmm1*8XA)Zah zneL6Ag7Q!FYeYlI)ss0wi$f(Z?Ybs$CDI{!Ry%azB=dKDr;7nBQLU2(pZCv8$MLB@ z`4lidi#wjpFYx|RI+XePO*Y?V1qBx8z(^l?iLUt9F@W%uBM&)UzqDOJMOemhy$7=P zJrndAXbEuULF1o5L%2!445Nl;4J|EYCB3T&%jf&r^VP67eh^#c$++^V@SOa2XN2itfOw3Mg|5-BG)c3efPlVRaFUiiEfjglCo8WFnGg~!nxD(Cw&kSgYZE~xm z{f_QZp-B6Lr0;sV(1cgwy{enWY$iG=Pf)e$+4h3CiUlR*Xls(0xS?{D7h zEDWXgmI)+ViI|{TPAa%Rd#SN|Zl2X`ScaQabIpnWI^ul}{@n}(ZDVCMi#8Ml^9^K- z`TO~Ql!WwOy%hX+gsHm+_K5E+T9w<+s1&?X$iXP3P<3XZ6b zw9gU2Celb+EOVlthGILz!_TJ~@*hVgU*X;1-7EfH_ExeSYIzJV)YW8m^Mae4(G48= zSrQJt-DZzO8@38~%*)mJOeSV;j=1YB-Mljx>u(Z}XK(!Nt6dsD1RfKK#||puGp?2w;|lza%1|PBu=!ZgmisDDD!ymnJJ#u3g4pGlX*o2OPdlyW0S9rh z63Brpz`6s!M(|n$5~Qq7JkqgqtL{i=vDMz|Y5xbSD?G(VWUd!?|m3jXA#7kbJVE!9HF49 zpJ^3bH5wKIV23YhG+0;N;o{F~%o559%dKPt6&3qvUkB>xtIKak`CG)v*SD=*>f7|w zdM^{>3}LW#zU2w&2}k|syhB3G&cX1CR|d%mhv_u9)qQ*UbE;|&#D!B88Ssi8@PmaY zNMzylLo0A>g8(Zlh)dCcW+>?^V%Bu2zT4zIDY)EPP@PeF5SRh`Id=zVh6l;7AFXpxSmtUMw&s+ISLowVQc7Voe~sINQ|j$0;k=`kab z+wVVRDn8=#B81|Tvf}bi6iir+_i}XGKy0T@zBoVQ9-FCgus~|xrG}1zY9H~9Par$# zD>7h>hwAgqs?-+CYKo9_UHIAQGJ;Umkt^L@!2ZwWegv(BseA2yTgQU1z@4D2GqrgO z*up<5tN5XHxbKVGGJ4q8?{I_bwUV=N=h+%GSWxBgsY#Lsa#oF+q!d)#UEL$yx>ToH z{Ys{tlKEV>a;uPTamSTkZt$DG%w`_fz!ARyF}Mvq_TO1!*#`nAv}PmydkgEbv;BO) z$vbp(;;a|?t2e>J7^d4H#x@YmL7gpYaE+|K`ArF6YCKad&=9^jGJ1!Mk}i;NAWhOR zPKYP4JU;cIpq9|516*egp$}9x;{|D&agis$T?}}2MLVyG@QnSEXcf(^va|?*%5c7# z$K&KVybWawU-}X>r-d`;AN3`5(-E&4Hnyz`_9bGO$ZNEE1e;h*pY(7Y_-X7HZX>9h}jZ%q;SZ1*uX=gbrSR!p~+WD zN=iGRLFU_z)xu9N1eJ&Frie#XX=(AeJ3v_gNzGfUo#t4xw%PP_{$>&O!>y>V3vY`h zVzY|KJ2@*J`p#tVCyR)JzevelWNAlOg;<1!O=UXR=8?yczGuT__Fw0=1A7AVlFbFt zh^>{{8@l-O{JPS*K1-%nK6|oWT!GfTDzu+81a614DXSqr&isB{+2PtQe-uQ$It7W5 zU?2UI{%7~BGm zF6)cJyAL1Yb^h>ZBhLG7!$Qog6BU{Cm8gCrvy^u;0t|P?Z+ti4Q8=>E%g!$K6THxq zy;l(XW-eQ@WZNulD4ptQk9~Ahrzta##1Px=ZlR5&0j<#Rd$HYw3vQ`9J3C;Uf`N%? zttdoIy_jFKDJdf(BPUm3At5Up5A7R3kOw2qJ!t6=0!2Vtr09rb64IZ_1W9`c8Yhk^4Cd+WxijZ9l%yQ3XKnY@h&Em{aKTs%2fIwHsGho4A3kavCs1~#aC8Z1${`C%hXAjbl+28|No{& ze?$_HgS+zkNUDB&2`}-T%XvnZDHl?T1la6iGBVvD|Bv*y0FB3-la6uc9wH*5*VnMN zqv**ETKSj~6%_5ooJ&Gc6~%U6p6yWhx3LP`;`Y5~7_>oYBK$-eHd4SA zn@kboV@M34H*DsM!fa0RU?7;Yz3!q)g;o(yQ}IYi!r)s`PrAmOZ%vwJVcWtERufmClNsz7Q&=AhuGjQM} zh}Y8En!I|$aiA21V5*tJ2}A1W237b}3Cj6v^MUC)Pxh}9XZ9V>;!yb=*(qj6`RU=$ zg1}khN?i@bsPHe?*a-7xAM(XDT-}dBdtx8y+(Ce!7&h{b?RwT zR9&;P>Vn(Pyi0@j^abi!S;ATmOsy@fuTI}(W4;TpgZuW{Rky)(8Bk^bFBx}Gxd3O> zmInDOWu&b!zljvG6nspy# z-I=A9+y3+Cs$_JH+~VmYZ5^mBg;b#K8L-+`uJ6g(P%Yg%;AVU=o?9uAQ`3<05-t5h z0jo&*dC6#7g&4MBfobX1&6|5RM-_1uz1$;A!O$2QtR{>5oejEQRzB03y_X92;9eu> zz-<$gqCT>+ZOL7uHCA6WN7t=MuR)$?)FExnlm2IiIjtfR-1t@Kfo+vy+N=@D8*rD4l;7LHx!G9yiqi zt6xwb$yLlsmyTYjsz3o{*>8+aj>Lp21N|=9x%bmz`93On+sit?b{0vbd1-a3#A;vZ z>^k04QijIpTMY2qlW)_@`83I9-EsXX=^{Eu4aL*W2jV&1pL&wofrVEV%AKKR<0E}0 zvs$&sH&X#53DL#-Wi*N2Ty}}FL-ePu?8(M6JbVfwhEhND@5Ox`NTKgqUc+lVdrE+k zFx9sqlIFI&gM}8Z>J=t|jYXUgn;K!@#Lqc9TtR&xnL35qF&Nq+Ij_;-)0)Br4x}nB z+#VI(L1|z2UmA!D~4WlJB&XdwgcTTm(Y4bq~myxjZHpJ z73P->S%zv3<-4U|$hw1SG2VZoDkYddKZz&vQTn}#qfFWePWJEdFM`*$uMKEms;+Nen;vi zHfYZ$KaMwT+q(-~4aquiyWPdfETvSGO-WrVU>bEGM__IFzTE-U=}_O8Rs& zd*A75Z7Gw_0G2*D*yNIQFUtyvO#qs~{gtb@K^M+|BE$2n>nm zll4bUQ>3r??sG){zC2>6-|Jb4Zm*ti=S1`&F#oCjkvxXIAbcKQq5r2T#R!4+4+#%D z@t^LKL==L5T1z@cKDKeME12-L+>>3^9g97E zTTgw-(tjfSlrM)IdHhFYXcY23O^aHw$Fo~He@0cDBLo&wI06eC9Bha7tJCG}O6J8} zYOBMSTfUzah0Sf!6MaM}z_-#_f7l!+RCKG8Si09*s#4P2|~rCpC2FW zbZ-~X`jyP--1>vMgq$Mdd$6U=p+2Ry%SA({h7--v?=XD-@Gc}Ez4gQcJzFcA+0|^f zVp*Ce*uiaaInHnSTV9l#i^K`OknD+UG!1-jn~T6PvBNcCC#G|e#l5F|zpmrsH)g(g z_xe8=I2C(mpI!a*;S%0EBsE@9*-SpL_Oc1dHO8d*mS^2bl%Mlp zG#t2N(MhI^u6TbX>V5)aFZ>d*^$a-@DVs&i)zV#yy(#mBMNX8Fx8fRsutjc4{$+*?Wm6o0|{&6$Xd9kq92;HgMl<;~g6gWp$)`TxJybJI0bbnpRB+BA7qb6_w^;t4g6#j^RQOLEge{XCL9$Zi zG*0OLxZ3esLt^7Aj6u<|A&1|LpT){--uXB28eT*fGYjpd4!^co(sVGY69Ntx+uGPA z=6Hgusp~|baDGBrwW$Iq-C?gkD(-P7rm35#lDFXz=n05)FqfGP)tA{DKAFPn`~A(y z!2!c0&7K7&4hg1!x!>f*)M4IwWZd^fyr)0CEOjNVDSaI_#@-MMOAz5@+1Hcwtk`$) zoE)OJ&k_Xw?iM^O*r(LxKbj}nSaHBH$p#@_X?l_7vBM#|O2Gl=Bry+ir2`TvW=Ws_lI~~zG!v4WEE0U}t}dlywr0~)TcA%d^t?CW-Bz}s zl+S`YO}MMg!76LPZ$&OVIOuKULGwM&q$r$OJK}7=+3?B~3Cyf4hHZ5u6v9!mnYE+Y zbbc-_VIE$#b_%cX?1f`^k}wPg?{$YKc-?&Z+@L#a+XnfbJ|}RTVw00>G5k@(^bnn! z?&36Sm#$K9bJj+d3BMJqWmUQGGSB?1sMe+`K38MGYcd zVJK7;$VyE7<>{DZYH03dQ_KiQTCL_N_|7{X9(5%w&3fzM1tmJ7kyK&MswwsTd%xAI z*BHJu^AA7tQs&9XXHfQ_naSchqg45P3HaS(b>!74mSp<9d45YSB!}Jg8DFIEH#Ngw z3ORY~uWIFIzsN4j6%xOl#_N-gX3GrbLo2_|9@UwkG`|X$lo^ct`RS%RzL*j(zAbt6 za}*E11&CR!mkz0^&73eiHi$ZKvx-wgUSv@T@2{w*Lgr_?{&=B7m=M5* z4CMI{rIMqkI7Q$;jo1FPIBS+djgv-0d&EFRX+xCdqr9-ZZ!9cElk{wn7_ zC=f3FrY+;Lo)y=rDQ(Q7_w@XU)hG^G2hS^;aZN9g8^!xKF4q&3-)rI-ZC@sKV?k&x zr(SG)_cmLC-S+xt5~H=1j6DT@67V#@Mx&^BTrgT@v}`;z6dyxMAZDlVHK@liXxYMj zu;TVpXmGxIYCVf08hzl_^mg`9UxWX&IW9)laru0sjM0+jQPH)dtT@Pdl!;=W{_VGZ zf9*dJ?*1cv5J0b26rc!dTU6i|eRO~8&S4Ov-=@9WRYtwloMg+2J-j>?`+7#lOEvu8 zE_Tz+X9<*dN&``K9|pB4$;*EU4ofu{lco&~3JRLAstwlChC_S7 z2qrh4UI(+hl>SV6GK;XtNa%Wme-Q=Gu&KuL^Ya6zbL>2(%Z`z?%0OZbQI7%wp~TL6 z3BJu23SNV2VOX^~1*_tTm>Gj>)>UICCc#(&l>H{ba(`Z}=6qkaZbzFj;@qW{7X{>} z!DU#lVsaVPf<5}iC(;}a+E5Akxg>*H2V^?%znH+SLQ2QUz{_eVCs{m6%{P)aJu|Dj z!Q9e4pUWBy!un?}<$mL(=)Lj-$>SWR!!B#u2haaB&s#A@Kw^rF$cXoEc=AQVguk&i zs4H&T8GI)wd*mv2f}hWuztwqK!Tt*Fg^VcjUNk*=>(NgBqW{6#QrT|cN3~#@^<~Ue zt-^GMFIVAMFOkRMB0f0nklxFhDxK`4S8-7q>*#_9Mvo3zbt1NhSm>~0*&kk(_I{SG zWa;>1jbrY7OIax!d}R9fp5*i2Vb4`Zb|lQ@+wl#x$wjk?24qbZ{J~JT2;jTQrAB)L zVq_LFSMQFpG31C5f+h~xcCLe-m-qJwzRJ5>t|mqpFo%CEEp6cl`4aXS#$`eQx&C7g z67-R;W5Sc^z4TQ-k~-H%If~TIuXSYHdP+=lz7l9*cl|SC)erhRs^wqpbT_P)yMOgOn>Vvvypme;&dV|%mRECMr)9dWx#(KT| zyUqp&Y!kCD`%itU)BAi6ws?p3LLqeK5X1O~h^O#~5Qz`5ub@Wa{xmHaFD@P=r*JTO z`K7^~uIw4cy^jSw`?iZka@-V59}BC=F@JqI#d~JR~ zLBad?4Pa*v+Sc(rR(dDv;57iwD6GuPGS~6Q$Xr0G3J^39i#Y=Kz+9z_K&l;&Y?Zlk za;E2{)f;$B982Q z2Bfr8-L*aBK2!%mn;vOcxQt3kPVNe}@F21b-jC`aSgZ2>y@W-iF0MegKp7*@!>h1OXQ?dWCWZ{?q2x)`*f# zpbx_--xj@q<5MR<7mxi0Qc6Ofom{|?O;AufS_V!w^V{8Ff_tZ=bPH>;=H`Nt`S;_L zz4W*j3wRWNb8kc@)o^vAZl&du>-~UJ=`DiN2hr15Xtj^ETU=Q9c++9#T}Hb;BiRj1 z%wGd({*xOsP~bC@!=k6|jOPEdEnFO1mBykCL3B8>UXcf~Z3F^Mc^wItyn~V-INj8b zIm*fg-UKP6$P#mF>*AZqOVztS(xf8I4dov{+0vb|9gbJ&`8bL%Wc_Z+Zy7ml6F-}{ zV*l~Mqn5;}p56q>X3Trl_8S^K)u60_|1_+29z14QnQ7qx1E!efCSEfQB}TQixtW$V zTg-c7!hR=*yS>YD&r1w{*Z*-IZIl$uLaF#MZ&^8QK69CES*5j0WM4DAINyOsxP4MU z<=u48{Z+~M66IBb75f@;9P}-B6@vOzaQ?KdCgrD4mzTf$(Xxhj2qiCj1g$<56huI< z7Nnt4fldwfd(4J!#Kb<`q+%;PcC7+$#E&;YfaShgmCMO4yt$Hr=vS}u#L1pq+J||a zr$k|mlwz*NcGr!K^oe+|ec8ECiyT}_ZOrt6jODkzb4#fFBgh=$y5H3FtK_QYBsMU8ov!8V@Xie{?qJw|Kpk2vIf_5t3}mU@q;vyT%loQZS>|hI|T@ z7!&l{AanEgxJ|6a4MBeT~+f3~lJ2%(aeCN_kNaZ+n&Xqx{Zgq?j<2K|x1GEnkZlq<|y= zWD)_@#G4DC36VMc)b6YZ(bR z6ar)LMu58OXpY*}r;+fWv@K}e_$dgdJ_Fxs8VPVUZ8Yn_q;E;(FkjoZXR={9%H)vz z6rT5Hnsh|!FOq`D)+l4y25FnPn|ffg@btPkhNy)|g+XToM7IQ(x0b2!|G~k*5QUHt zwAN}GtPvrjb>9D}kReOnQgN-)?Ca^U@X%uD&&)NmZoH5$A8+3K)QG-5?V8voK1dp% z%1@3QHlu=Tfrmqt76?y1^-b{4r*#%5b`USXcGB6Q@Ph&hGqW-iVl@o z>TL++XNQv(OuX+PI;I7J+EH&T@7|%;UtsnzIUN2URFhuh{{wi?|0DSWfJ%=^L7c1B ze44LJNm)rYWAIH$boZMwPeVs6* z@JCdF?{sc0s8_DY#qz*x!9eb=D1yYV-;)TLcv2fSxe?J&*{}Puk{PyaKNfcgHr?H> zX3w1$P(g5BJTFMr{L>}}dpw2yBqsuG9MT5QE&EzMb_)Nknj~oSr zD^Rk#C!(*P+M=g`{0k_#e*S#po(NR!VZVqR!hd9@vPsA=@jnH&HzwL65ZsHPMGewb zy;9nxP?Dg?@*?Toutz%Y!L*c?l|9_qF&@bshuAb(YMf%vl87l2#WoMBu*=dffF=RA zgRt`o{T9ScuwZ@;A33=K@a*uv%?$PkYT&$AW`oiJnDp1MuV1t6uReZ+A|zy)%GPP-w5W9EC`l+KBH^^yW{8rl4QVV1ku{7-%vj(1+wXec z>%Fe`djEO+_f1%;wIjRpNX^w?JzjGVG6yer*J$=gSHO3&3xBmxy>8U zwGE#d2gBL~rm@ImvW?-2TeBaYmPW$6$uUc?;b_zQ+}T##_e^P9YDRrQ*hKK1n!Mjj z{-~(1wBo8;5#(;*{6Cu8j86ZZ;LW$vVV)D(~w-^w+Jt^~xP!qVV>h0Sh zI7%#3)MaSAyT7`wqkQ^wE@*&HxVnA<;p$+2e@3vvt$AO%y5`}H_vHAWfNHn}XGi4H zRl>;8ite|N1^_%fLv&dG4Wi@q`B29I<6S+zS|*qP$U zA!_e0GVXi}C`oMt*1NZ{& zD==mC1Qx*QA0MNz&S2Nqx)kU_9TDgrH2Da#l@%3W(ksDxsmaUV4mQ|KMKS-e290J; zZVxWRL_z~d6r;)lFp@xsi9qAV?WvC=BqO8$+A<$%$$HcmklzUT+CXJzJ7@zn#11h` z{o0;^r_O00GLiu*?3lPXkgV-!GUS}D9E5O5ZS<|oG z-5TsG>i4aaOh7Cy$M9(_a`_S(xg2zThjW{yWmR{aIveZ2Kl!BID*^6#i>&b?Rn@_c zjyiN=u`M5QqA|4xw$k%WQc^gqr*CYTZvKX?+Vp6S-u1Er@&Mo;`54DxoMW?Rf0{`f zzRC3P@p({r=yNEAbsn~k+o^M@a5wfoHIY(3vE|U#KRYjnq=$olDaR=jOq|eW*j{;! z8yB~l?b(n`l|E#1hQr~^rCw+$*+uLxM=m@(oHw(p19MJ>hK3VhM(oT#FI;Glz{Hv4 zUfg;x$6R!dnc9(+D_n@q(!tQMFk-nNya9T?#m*zJ8yPC*GH-8PyP1l4D0v20iT}{j=)3MCz75$y7qc z8+xE8Uijt`bDbdV`!+VFW*W{*?~P2a2kU-zb~dn2n6N||(UC#UZRSJ&3Iwo>)VdZB zlzK_ePPYpWpN8xQ&B}`iL5`#^zbNVNAm`wb0h#uiGza&TypP|Bh6kAF^BVdfu2@UH9Y>>4Fnx4#rv|=3Mi z-@;%p@B#-12WX!VgrP`U(-LIT0>B^dyKmpe?ETLZS)b?PL$?_sfrd;r-IaGeG0`=Z znShsGR5dnOw{G{qzQ}<8>NgAi*FN+Ah|B+%X7zu6lUsp~j*bL#C6wtgbmpNSACFY2 zIAR8zy#Yu$y(}^o#a7dZa6nA4x_1HayCx{Z?|tnK{N2XveG|60mBwh=!~7=yq>BEhLoWYypi)%+8cIHFBV+UyUtkvE7HKN5Xt zZxIBmAM=JWjRrrKix3tQ6GO-gC4w6@RP1MId9G>ya6}h=3E^m%AEHru0{Ypu=xLDJ z=}2SxDo^fVmjpdhQ&LM^N_xN!T_Ada2t2H$-E zfiORyO@N*CU?L?rM;}P`rgN3Fre}PG*oA)|ek_F7yIZG=i-{p@l#7O;E*U#M?(gr9 z-$1+vLH*Lji#K%+LW?$CE*X!>Rfk@J6unO94Qr1bN-0KW&&^U$QZhVL zaRmt-1PCE=Jc^UHbH@%|m?35aY<18YF^Al(Bl)(VA9<=DB|~ka-!cq>cV;OiFmo}) zn7p2_4L0bf4IwC9i&I2pt6M2VrTAS%DD`l6UwmrHLiNPm2d$m}z5WIXob?IJRHGuu z2UV#@w4Ijjl+cQMGmkwe@0d#_%P~U1=ME<2`rg39kMcu?K0%)-#vU__jr>y~H4@+4 ziTkOEnv%~%wuCxDX=y224wb7BmJxjUnoUPB%Z`b12#!-*n{F+^=0gz0F5ndE<^>3W z*iAy~go-=SV8yk|a!rVjca{Xz)z;2gpnRRnUCz#%nVx_|F(0OiY1A995p0Y>u$%ko zWJZG~A_|mp3bAmgTC7{QuC}g@LZN*5T!0XXQdeG2CmhiXsDbUzp)>E?r@+lx|6`f!B52`39j1uwx9m(0s_SVZ_AV6&LO`sq zW%mG))r`~6jMv{u7ywYFZQvvrw2+rG$nfC3wX_p9CbQ^7-WtV!kKB1Vy$LyQds{15 z?5Jk8wzl9tuBn7$LitO90bOsNUv_%*v(|SnPfpKRBJf|ecU?WYyQZc_l#IhwwE{&a zWXOzjKvH*8QWn1n+^7t`F3pd}MDi@hoLO$72}t(RH8fItmghg0p#0`&9wkM{ zFI*@><>mWOtJ}0SKrqsFh=)AJiX)ScHXdz2FweB}u3Su6xUiFHa`5Fv{QF{;_sQl_ zvL@oq59!l%le-5w-{2x#YjS>(gn{ zOiRnW#(^T84g4HDH7L}5a!LvZ3o?Z6U+CA?x@f8$j6Q$c4wtLZN5jyl4%&*OevzW8>NNa@GEy@l)4(_(b1b4*8*4FwSHfQxAvvQL zPX@G8eH{T-CMWY9A056L`Wn{S;$Nt641M}yu!jT1g(5}X zz~kh$-P>SiQdq>zYK7*f$Sp;RQ&%|E_UL!+8jtPSzIvaOxfg6*0{CO5BNcVI0&;pEhW>~j(+ zwu))^zM_B$g{NP;T1H-88*k$ufLanxJVy?d{nbGX$5RiK-KTD&fa6I<`$(9YmzM|F zHkV4oHomJ{00RnOh{gE2&3J5N(x?_ID2V^L{I_C9a3P#viLjf#OtNo;J5P2aKtbHvlE$!MG=h5CekMheZZlS~GDf*(|yxvP|iiOocFx5#ux=XdbwM1!t zPle$1W6tqz5-DOF`qdCk$4s}fgnZmlTcH~4)f()Isw!^Mot4uwk%S0}7VorI4MU5) z5xUm*TT_H@yhKO3F$H*C_sDKvUtbiT;KQa~ym*1I1?L$K5l%=Fuq-VrGJPj6;Sn+~ z3Wa`;8nxL7#f(2-u1pG}_G11feULvg)6Fs14BJFD8 zJ%zdkX#36WLke54{CPJu==7}8qi}~%U&D`nd}e8-dp>sKE+$S(1Oj`q4#-%NBdq)j zTV}-NrjY2WcFJjZG==lAvvemqo)N=c@yDj&sx6{bjRDOO z-gs74?NgqA2TlUCEIdRh-|;6tN+e&7VAPaQ^T9xaG`0^x6+)G8OIWrC4mqEZUQ0RZY*3$=Qc(((*>z1^#N@{P|bu z1QEKP1t+9cio}YC;o}UAi^VbD3sqBdD;2^8@?P zo1PwZ_Ve?r+zXq!8W-m{Qie0@+@}S%IfY^dh8t9W)j7Z}?^tc5zz6P0ri(r}AFJ#j z5R1X|v!7&byT-a!L>mQAl=_ERvu4d=%K_EV^Xr(2Mys=yC@DE_>sb*KO(CJ1mhMMIMP?2v1iYo=Rkm^rM`m0h)Z4!5zENS1GkFr;E{yqH4^0* zkOQl--noUQf}nuv8{-_}A9N|6;07>><2SoRF@BUplJc=K`BZYw?Y3(N|l2DUqWv8n|M9iRxAVVeO7Xa3iqi{3p)2Sh2>I;1P`ng0f zDn*%!TV;w@EVMM3XfBQ`cp^d^O9e~v2yg8Ue zSn%*P9ikqoh+!J0twpHA6azmkfOZwpJlSkcj4W*qBI-D($D&XS9S{Q@9UU624E-ov ze*w;e660aGK-Ba|q?niT|2gIC{r^2j&=w&jwX#a@!Z$Y&c7SAM&NO?-`2COn0G$mP Apa1{> diff --git a/docs/03-data_management.md b/docs/03-data_management.md index 9ca8c41..54573c7 100644 --- a/docs/03-data_management.md +++ b/docs/03-data_management.md @@ -8,12 +8,149 @@ title: Clinical Data Management ## Learning Objectives +Learning Objectives: Learning Objectives: 1. Understand the clinical data handling process and how to choose clinical data handling tools, 2. Explain privacy considerations for clinical data, 3. List major regulations and regulatory bodies for clinical data, 4. Describe different types of documents used when working with clinical data + ## Clinical data handling tools -## Importance of keeping clinical data safe +Clinical research relies heavily on the collection, processing, analysis, and management of data. Efficient and effective data handling is crucial to ensure the integrity, reliability, and validity of clinical trials and studies. Selecting the right tools for handling clinical data is a critical step in the research process. This section defines data handling in the context of clinical research and contrasts it with data quality, discusses the importance of privacy, provides a comprehensive guide on how to find suitable clinical data handling tools with a focus on open-source options, and emphasizes the importance of testing data handling methods using dummy data in the preliminary phases of a clinical trial. + +### Clinical data handling + +**Clinical data handling** involves the processes and tools used to collect, manage, store, process, and share clinical data in a way that maintains its integrity, security, and usability. It encompasses a range of activities, including: + +- Data Collection: Gathering data from various sources such as case report forms (CRFs), electronic health records (EHRs), patient surveys, and laboratory results. +- Data Storage and Management: Safely storing data in databases or data warehouses and managing access to ensure that only authorized personnel can interact with the data. +- Data Cleaning and Processing: Ensuring data consistency, accuracy, and completeness by detecting and rectifying errors, missing values, and inconsistencies. +- Data Analysis: Using statistical and analytical tools to generate insights from the data that can inform clinical decisions and study outcomes. +- Data Sharing and Reporting: Providing access to data for collaborators, regulators, or stakeholders in a secure and controlled manner and generating reports that summarize findings. + +The goal of effective data handling is to ensure that data is accurate, reliable, and available when needed, while also protecting patient privacy and complying with regulatory standards. + +#### Data quality vs data handling + +Data quality and data handling are closely related but distinct concepts in clinical research. + +- **Data Quality**: Refers to the accuracy, completeness, consistency, and reliability of data. High-quality data is essential for producing valid and reliable research outcomes. It involves processes such as data validation, error checking, and quality control measures. +- **Data Handling**: Encompasses the broader scope of managing the data lifecycle, from collection and storage to processing, analysis, and sharing. While data quality is a component of data handling, the latter also involves aspects like data security, access management, and regulatory compliance. + +Both data quality and data handling are critical for ensuring the integrity of clinical research. Data handling tools must support high data quality standards through features like automated data validation checks, error reporting, and data cleaning functions. + +### Selecting clinical data handling tools + +Choosing the right tools for clinical data handling depends on several factors, including the study's size, complexity, data types, regulatory requirements, and budget. Key considerations include: + +- **Compatibility and Integration**: The tool should be compatible with existing data systems and workflows, allowing for seamless integration of data from multiple sources. +- **Scalability**: The tool should be able to handle the volume of data expected in the study and scale as the study progresses. +- **User-Friendliness**: A user-friendly interface can reduce the learning curve for researchers and data managers, improving efficiency and reducing errors. +- **Regulatory Compliance**: The tool should comply with relevant data privacy and security regulations to protect patient data. +- **Support and Documentation**: Adequate support, including user manuals, tutorials, and customer service, is essential for troubleshooting and maximizing the tool's potential. + +Research teams should conduct thorough evaluations, including reading reviews, seeking recommendations, and conducting pilot tests, to identify the best tool for their specific needs. + +#### Open source options + +Open-source tools provide a cost-effective and flexible alternative to proprietary software for handling clinical data. These tools are often developed and maintained by vibrant communities and can be customized to fit specific research needs. + +Advantages of open-source tools include: + +- They are typically **free to use** +- They provide **transparency** by allowing users to review the codebase for security and compliance +- They can be **modified** to meet unique requirements +- They often have strong **community support**, which can be valuable for troubleshooting and enhancing functionalities. + +Examples of popular open-source data handling tools include: + +- **REDCap**: A secure, web-based application for building and managing online surveys and databases, widely used in clinical research for data collection and management. +- **OpenClinica**: An open-source clinical trial software that supports data capture, management, and analysis, compliant with regulatory standards like FDA requirements for electronic data (FDA 21 CFR Part 11). +- **KNIME**: A data analytics, reporting, and integration platform that can be used for data cleaning, processing, and analysis, with extensive capabilities for machine learning and data visualization. +- **R and Python**: These programming languages offer powerful libraries and packages (such as the `tidyverse` for R, and `pandas` and `NumPy` for Python) that support a wide range of data handling and analysis tasks. + +Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform that offers some basic statistical and data visualization tools, and offers APIs to allow the real-time analytics in other platforms (R, C#, cURL, etc.). However, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued [guidance](https://www.r-project.org/doc/R-FDA.pdf) on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR. + +While open-source tools are beneficial, it is essential to ensure they are secure, well-maintained, and compliant with relevant privacy and regulatory standards. + +When moving between multiple data handling tools, it is important to consider interoperability. As an example, REDCap and R play very nicely together via API tokens, but this may not be true of all electronic data capture systems or databases. The sharing, integration and redistribution of data between systems - whether internally or externally - needs to be carefully planned, tested and documented. + +### Testing data handling tools + +Before implementing data handling methods in an actual clinical trial, it is crucial to run tests using **dummy data**. This step ensures that the data handling process is robust, efficient, and free from errors across different conditions, without risking sensitive patient information. + +- **Creating Realistic Dummy Data**: Dummy data should mimic the real data in terms of format, structure, and complexity. It should include various scenarios (e.g., missing data, outliers, data entry errors) to test the system's error-handling capabilities. +- **Simulating the Full Data Handling Workflow**: The process should involve every stage of data handling, from data collection and entry to storage, cleaning, analysis, and reporting. This comprehensive simulation helps identify potential issues early, such as data loss, security vulnerabilities, or errors in data processing algorithms. +- **Refining and Optimizing Data Handling Methods**: Based on the findings from dummy data tests, researchers can refine their data handling protocols, adjust tools and settings, and optimize workflows to ensure smooth operations when real data is introduced. + +Testing with dummy data provides an additional layer of quality assurance and helps build confidence in the data handling process before the clinical trial begins. + +### Conclusion + +Effective clinical data handling is fundamental to the success of clinical trials and studies. Selecting the right tools requires careful consideration of privacy, regulatory compliance, open-source versus proprietary options, and the specific needs of the study. Understanding the difference between data quality and data handling is crucial, as both are essential for ensuring the validity and reliability of research outcomes. Running preliminary tests using dummy data is a critical step in validating data handling methods, ensuring that the chosen tools and processes are robust, secure, and efficient. By following these guidelines, researchers can enhance data management practices, protect patient privacy, and achieve meaningful and reliable clinical research outcomes. + +## Privacy considerations for clinical data + +Privacy is a fundamental consideration in clinical data management, given the sensitive nature of the information involved. Clinical data often contains personally identifiable information (PII) or personal health information (PHI), which must be protected to comply with privacy regulations such as the [Health Insurance Portability and Accountability Act](https://www.hhs.gov/hipaa/for-professionals/privacy/laws-regulations/index.html) (HIPAA) in the United States, the [General Data Protection Regulation](https://gdpr.eu/what-is-gdpr/) (GDPR) in the European Union, and other regional laws. More information about PII and PHI can be found in [this course](https://hutchdatascience.org/Ethical_Data_Handling_for_Cancer_Research/data-privacy.html) about ethical data handling. + +- **Data Anonymization and De-identification**: One of the primary methods to protect privacy is to anonymize or de-identify data, removing or encrypting any information that could directly or indirectly identify an individual. +- **Privacy-Preserving Record Linkages**: In scenarios where data from multiple sources need to be linked without compromising individual privacy, privacy-preserving record linkage techniques are essential. These methods enable the integration of datasets by matching records in a way that minimizes the risk of re-identification. Techniques such as secure multi-party computation, homomorphic encryption, and differential privacy can be employed to ensure that the linkage process itself does not expose sensitive information. +- **Access Control and Encryption**: Tools should support robust access controls, ensuring that only authorized users have access to the data. Encryption should be used for both data at rest and data in transit to prevent unauthorized access. +- **Compliance and Auditing**: Tools should facilitate compliance with regulatory standards and provide auditing capabilities to track data access and usage. + +When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research. ## Government regulators +Clinical data is governed by several different types of regulations. In this section, we will review some of the major regulatory frameworks and organizations. + +### Health Insurance Portability and Accountability Act (HIPAA) + +The [Health Insurance Portability and Accountability Act](https://www.hhs.gov/hipaa/for-professionals/privacy/laws-regulations/index.html) (HIPAA), regulated by the U.S. Department of Health and Human Services (HHS), establishes national standards to protect individuals’ medical records and other personal health information. It applies to health plans, healthcare clearinghouses, and healthcare providers that conduct certain healthcare transactions electronically. The HIPAA Privacy Rule requires appropriate safeguards to protect the privacy of protected health information (PHI) and sets limits on the uses and disclosures of such information without patient authorization. + +### Food and Drug Administration (FDA) + +The U.S. [Food and Drug Administration](https://www.fda.gov/regulatory-information/fda-rules-and-regulations) (FDA) regulates the safety, efficacy, and security of human and veterinary drugs, biological products, medical devices, food, cosmetics, and products that emit radiation. The FDA’s regulations ensure that clinical trials are conducted ethically and that data collected is reliable and accurate. This includes oversight of clinical trial protocols, informed consent, and reporting of adverse events. + +### General Data Protection Regulation (GDPR) + +The [General Data Protection Regulation](https://gdpr.eu/what-is-gdpr/) (GDPR), governed by the European Union (EU), is a comprehensive data protection law that governs the collection, processing, storage, and transfer of personal data within the EU. It aims to enhance individuals’ control over their personal data and simplify the regulatory environment for international business. GDPR applies to any organization that processes the personal data of EU residents, regardless of where the organization is based. Key provisions include the right to be forgotten, data portability, and mandatory breach notifications. + +### Honest brokers + +Honest brokers act as neutral intermediaries between the data source and researchers, typically regulated by Institutional Review Boards (IRBs) or equivalent bodies. They are responsible for de-identifying data to ensure that researchers cannot trace the data back to individual patients. Honest brokers must complete specific training, such as Collaborative Institutional Training Initiative (CITI) Research Ethics and HIPAA training, before accessing data. They play a crucial role in maintaining the confidentiality and integrity of clinical data. + ## Documentation +In data management, several types of documentation are frequently encountered, each serving a specific purpose throughout the lifecycle of a research project. These documents are crucial for ensuring data integrity, regulatory compliance, and effective project management. + +### Data Management Plan + +A Data Management Plan (DMP) outlines how data will be handled during and after a research project. Typically created at the beginning of a project, it is stored in project documentation repositories or institutional databases. Often required by funding agencies, a DMP may include plans for testing with dummy data to ensure data integrity. The importance of a DMP lies in its role in planning and managing data throughout the project, ensuring that data is handled consistently and [responsibly](https://www.tableau.com/learn/articles/what-is-data-management). + +### Statistical Analysis Plan + +The Statistical Analysis Plan (SAP) details the statistical methods and analyses to be performed on the collected data. Developed before data analysis begins, it is stored with project documentation or in electronic lab notebooks. Essential for clinical trials and other research requiring rigorous statistical analysis, the SAP may involve testing statistical methods on dummy data. The SAP is crucial for maintaining the integrity and reproducibility of statistical [analyses](https://www.questionpro.com/blog/data-documentation/). + +### Standard Operating Procedures + +Standard Operating Procedures (SOPs) provide detailed instructions on how to perform specific tasks or processes. Used throughout the project lifecycle, they are stored in organizational repositories or document management systems. SOPs are often required for regulatory compliance and may include procedures for testing with dummy data. The importance of SOPs lies in their ability to standardize processes, ensuring consistency and compliance with regulatory [requirements](https://managing-qualitative-data.org/modules/2/a/). + +### Data Use Agreements + +Data Use Agreements (DUAs) define the terms and conditions for data sharing and use. Encountered before data is shared with external parties, they are stored in legal or administrative offices. DUAs are required when sharing data with external collaborators but are typically not directly related to dummy data. DUAs are important for protecting data privacy and ensuring that data is used [appropriately](https://www.ibm.com/topics/data-compliance). + +### Data Sharing Agreements + +Data Sharing Agreements (DSAs) specify the terms for sharing data between organizations. Encountered prior to data sharing, they are stored in legal or administrative offices. DSAs are necessary for formalizing data sharing arrangements and generally do not relate to dummy data. The importance of DSAs lies in their role in facilitating collaboration while protecting data integrity and compliance with legal [requirements](https://atlan.com/know/data-governance/data-compliance-management/). + +### Documentation across the span of a project + +These documents are encountered at various stages of a project, from planning (DMP, SAP) to execution (SOP) and data sharing (DUA, DSA). They are typically stored in project documentation repositories, institutional databases, or document management systems, with access restricted to authorized personnel. Not all documents are required for every project; their necessity depends on the project’s scope, regulatory requirements, and institutional policies. Some documents, like the DMP, SAP, and SOP, may include provisions for testing with dummy data to ensure data integrity and validate processes. + +To ensure compliance with these documents, organizations should implement robust data governance frameworks that include regular audits, training programs, and clear policies and [procedures](https://teamhub.com/blog/document-management-compliance-2/). Standardized templates for these documents are often available from funding agencies, regulatory bodies, or institutional guidelines, helping to ensure consistency and compliance with best practices. It’s best to ask before starting any document to ensure the correct format is used. + +For further reading, consider the following resources: + +- Guide to Clinical Data Management Procedures (GCDMP): GCDMP +- Books by Suzanne Prokscha: Writing and Managing SOPs for GCP and Practical Guide to Clinical Data Management + +Additionally, remember the insightful quote by [Damian Conway](https://www.questionpro.com/blog/data-documentation/): “Documentation is a love letter that you write to your future self.” This is an invaluable tidbit to keep in mind throughout the lifecycle of a study. + ## Conclusion diff --git a/docs/03-data_management_files/figure-html/1ivDTcLjb2078O0GemkSeCgC1jmxk4fMsiFQaPaer9mQ_g3edc229d226_1_0.png b/docs/03-data_management_files/figure-html/1ivDTcLjb2078O0GemkSeCgC1jmxk4fMsiFQaPaer9mQ_g3edc229d226_1_0.png new file mode 100644 index 0000000000000000000000000000000000000000..7110e241e32f6c05c6544df687bfb10690af02f8 GIT binary patch literal 66775 zcmeFZXCT(^8$O&ndvB4O$X0gQ-1go|WJI#Fqs(qI$yPSEk(D%%krm1a-N;@k3zQ6zP`+MF!Z=QGmcj7ay&vjkrc^>C+99MD1hFVnQY~&|SoS@RtRyRFy;$-`Y z69{%v1pMZevhleSC+?omQCBg)ZM&XF)_lqAcejzLX_UaNGyxvRkR_(a1ibh8bOpU4 ziA~6c?6Tu@Z$q~eV$jRn1}-0O4p;psZQS}Zne_Wg;D+BX#osf2v71Ms%0DXV z#%zNNTE=bHUc(O?sAO7IQm!akdT|2zyZ`x-gIo%J1HV?npcwxC_4CR9`>X#SW+D0Cj9ZFr*BzT)aP?nTOLGYvjO6wH)_hzz^1{aV+Nb(ho^Sr^F! zR9+7GauhwX^r6(v0%>{i!?914yX=X@Gur@iRZ<-Saax!He>Mr{!Q7v5wO;dciG0q~1);yP zCfg&wwf-J7R6q3l)QBAf4QU349GTD1qoCttr_t)0RR?|)%N8~Ls6kFAl#LPxZ;*PKV$ z8CX|ZJ#=ERTA30fPX8S&9x)XgAyds+ZAW(bfU29*uO?sK82PQ*_`QQd_;S?*Ooj+` zo_t`~&tCp-O>5O#gYq8?*+_H!24#;9cH~13YFi?|m%Kooyz=SOn@Fuh_C~LHC3%sN zAEOp_z0oJv#-G&}pFI)!pvtd_l*?8B0TKRt`iwDTCa=8aj|=`BYJDRuy!IvnDIZ|{ z(lIViE+viMXb`^P*4h}0wKuwMqmxECWbeCuaHLpC3(8rLP1>`nY_?h|X!*^ClM4aE#>N;k zCb)s;Y?=@PsTFH@K>!R-H1ke4Ato~15TG=vE+Rh@8t~= z5y@RDgzPLKQNatbh5Lv`81(d}@e7-mW26zdnZ}yD9-8h%B<#lgEc+U#fpv8e0c-_1 zQ8sOAijKzLc}Fn1s0RI7uLvZ-jXllD>rgH;%nMq7(cSL>H@frdlK=Y-Vf;I;q=_I} z?rb@&?-nO1eI0(R_ie5vmWETwYpywxgl1~>aBJKy{Jd@R-qr`@Z}V+!&7R8MpW4Y7 z_&g?_zcFNE^;_=GTKW!uoc4KhO;xo6_R(H1f5Ejt<-_lhNF=4_RIScM60Z4~KX45M zO`Ih5iG?Wp#&6g;u!tI=wG+mH_1C+p<>drre05~Qk3-}K?Pr^A$=veI3fjh@Bh^2X zmThVF{P@zHA!?`O_OV0Dq_VanShFhvvm ze(~&j0vxH2=qynlt?H1u`!^$FoVtlv&8U zhw{(S?)>ztn^#M;=?3};GgkA1H&s}c7JJi{FNf_f=Xh}huZt_TFw6QatGh-?zOm|m z`qAHq(&@IY%f$1-&E5OL)!zfG+b|3c%&FtH!2^c&V#3WRubGAyHm?IRYyz_Q#NLQh zUV1|OLh07m^u~wMFv>CZGY0{+{Jx@BYu-IfpLtI%qDYP!t-LhA;1bbc+~|AbUhyUE zbl;`E$!T>|gI|ODxQKz@`}6fzUVLlbN_+?TeB%C2^uErJ^nA!j;>Ekt%CN3bMm}!y zHTtboo0Scw3!1GDq@Jw5cu*@-&rv|zmm$h2gz{fJ95abZ_AZIWNBVWGDbnqsVl5~< zH~exxL^`~^Z;&n3JIGBL&myX+=;=(yAw`y)>3y^0bUjKfuF9e-s)vZ}*PYeJ8EiQ* zybv@lnQ4nS9f&e6E`km5buGyspWdDJ@=xm2Kn!+LNzYt!W+C z+l5^w;WOFU;gJ0m-Vv#hZ!dJl77oh#E@A~}bkhaIZ+-n((ip=lhc;{po+SJlvcHru zJsG`_?Osi`|764zc?j zvLNh^!9i{Xfvlai@f9s2G96i09()csl|qb2Xu9B-U=3FZv#VZw_w;K0FODx!SERh= z6s#E{UXjSg9W`NV~Abp{|!1JhK^BMRJKfRo4&&3N(mtCo)+qas=CLQEL^SBKvwCO+T{+f$Le^J~(1!zGqg(h0*~+RNxK zyC~%NA%FIM#B%HJ&DQ~%BAQ7Y8%ixQ^{MB3()g2@gfa3+d!z;_nbs5;;2t^Vl2<5E z%a=Ui$A_8*r*Gp5V>J`l>YP7t7(`|VzP-<6oCS4V@&*)1#i?2Tf{d4cly7dC=eCK)&5{Ypq}O25Qfn@+1j$?!}vIA4w2^ z3FvX~{8i;4W5r(+*UxG`xYFp$TS5>=!=ca+vj1~;KD^Ci@@2OEZP)`FRcdr~C`d@N z?B6@1Z!bi%@EPSd+jkP1i1X|n|NbU_>#KQ#x4i+gH+#^J&u;_93T1p3dmymAJWqQM z3K^amynCl`yvkA%5;??T_EEu9oqMN;g9x^i+Mz7K|AEVB<-<*)JZX73Iz5c^%_+w@ zI`2|yq+EZ#;&iV41JhV;^Om4nPoCW~+YtYFevipzw}o;fp|_<9hWPfYSbI;-1^1XD zmrWuOdHas2gPpajew5x}5Bfb!_vr0kA`&9UNr-Ude8I7%m7E^#_w| z+{=ogr6+O~!f$BCw93ClttV?a{Ag!f)T*|}(D822AE@8roE7yqr^gJ_Nv+LHi^%RA ztd^yPJ|iOYe7$jTfVBhJ8}CwzdNR)a+^#JwrE1-sTxu%#C>ZKdz2~f3*+mq?jeI%w zn++C6`b7-52qKEhRaZ=o6AQ-L!o!n{V$YDX-K0M3ysa=y@m)wD-4pSA7xw7HYAWKz zMH0G!tdrt&H~>KvfvQM{-(Pd$WUEL;$)qnklS|1_O3&RcK1A}e>wY~NWcMU-ccN=@ zUI^gXqlbxY_FhnxPeK2-^~3`?0m0wDHfolCVdmo_Ye!fyN$A1IMzXFc!*p#o3aQ(N zKe{g%Fdn_Ab07bj7ybKfV!)j1MdU8*_^s+)?3&mUq*%zFx4a4WvM}pIp-ZW@Ez%=p zMp_lGAtW|PRqurlTj7DH;MpBQwuE}wRIA6j52K<8c-s$so12Rn>84t$JB(NECHn5@V6Huvby_1L+0rnSkGlK+x?$Q8;l}F z3@(qo7dpHz%i9RVP$S)_?}UASLlSVmgz}VF`^1NNFU#oQAMX*5wZqMStxsMOb>PN+ z=}sC~i>R0$dT~;D?A-7I--B$#)9YEZo7xA4Dl4OQ#(UwjPAvuaxC~e1QkuLQ`4vk zZ@7?{VTf|2WwP=H8T?FA0PhDi4=^{P!uJ+pvn6gMP+vzbx zh2%xjrcRY4;p<{C??b|at~dq@YoBPx9Jqfz?Dv3S3cHj0ig+a>(eibNZ-Trk87LZd z8LZ(#94{8Cupf`4ih;i`(d{n@((AaA5_L?A#TCDrTWgm?Q8a2(&39_@xKjc)rdDmM z+l~(emSggjLZ=jH(5J|1U3tE>?B*kpX8kk_)NE2ci2>de{)}4DB9Hea%DdP1Nr~8Y ze?l+|_L{ERzT44wgOxJzclxxnV0&yftcl^*&&2dXPIJ%D8%FLEZXCg#U%(^&UG62e-`%8h#kD0=bb}alp-_j?XVy{BV`j)_F>RZy(H$3CTnmqFm zyYtg=fXA*iWa0ASJ!8y3a3eH)0se$M!JRd+p_QmyX{~*nwB8}OM{t(hHijw^7uudP@0foB0%|77h4MIjGXEHRrwk4^bDxKt% ztF<%L!4%y7Og1p4I7P<=;hW5kr7om;Rb2cSl3$QVZ{lh;;;4f{bOM5eTC(_dRV6uIjOgK*fwfcy8vIcq-APYhAhk=Nc46}4-fNlc_1;8|?(sNTT2C#H$Q zB1Y;cR}MGba+jB5Y`A(WO)2zXV(1-_BA~qxgF^8keUaDe5>LD?!rCgHU?gs%=_BP5 zulQas~ zw6so4$!>*8K>J}r;on2D)+-UT?J}?c&HIaDq;2@oErTyDeOF#)O8Q*q#^Nrz#0=%3 zd+vO#DD$@q{r#)17^#m|`n_KB=WxqNUZX41Xb}ppqn5n709L#$WA=kRY_if9y@uOK z*-{KHTg5^goND5Y0HS$A=bG=#zPHPKZTL$(o#fyZ@AH^8>y6=?Lf`tr&a;}T~M%9w(jhVT5RB4&_WvdPX)118Do)o?4_uBJR^}v zdDN_3ZQZ!ufM2AcP%|vM4YhR4%|yGHoMk#Ly$q4jDoCo!i|gJaLJmaDTDv<#TAY(Z z)TO8lp-X*MLeUZOukw1AcA>ahMl z>-0L%3np;+t!h?naT*`#?jARN|D??3&M}r`Y~rru(5tjAQsBvtKy@(gZWNOhKXnCa5vsRa?OU0W7*$t&GsUTv9MX`1N^6%Vg`~b zqjE<)igrmozbCKB<W;W`shUO zeP?OcGuXL;C!|_D2%ADxB7-e_S3MLa6SDW*iHQMyQ45=vXO~a^Yh&8Em4MZA*4hjM z5X(vH_rkNb zu6q!qj~gnEZ8^QNB~5!{YfIX*s|-lBCdXd-;9ym4@;wVoq`%9G5pf{yBdyEUt#4Q8 zeu{5H6mK@IiO5%Qak%?<4>4jio@y^sR*YI4Sg5YVNb<=PM%Brb+Xy4=6)QC zpN9u0X8+1#QZ-Cgman~FyQg#2L#zU&RX-Pv@5=`&foKY-RpUY2Y#BerP-WTdBuL(< z?o*|G9eq)`lHPZN?oBQ`=VrXDSIf?$(l{ouYr)cS9y1L;_E!o3)s?p-#6u6UlV+q8 z_j!n&tIL4?KXX2bpYx1!mPL2QDlfTTKIi>R=`~PC4fi%leJz3C0}dbID*$~_w$Q$o##3)uZ=!&*80sw zKOL`*G5ccs%9g>pHKcKq`*Oww9$?AdlyF`m+h6iqtgtKFOoLYE{=p2QUC2Fe;dR^5LLkeZQ-(R{uh8fpWN873#reu59?(8G-9C!BCzzAP@Hs zYBns{ans?Tgt@cX6%fU-o9R=qhsb3w#5hIVo%ET_m9py ztwV=?3!NwYtaJs#-U#p*(S2$>uk-%SH1F+gIZztglxyH{OwZ~QapIOD>YeK11 zf1dn!qM`V*`|8c9+84!&g(wQwvsh!1qjLH7=o&Nim_PwtjKi%Cd}0)Cj0Iu=1M##E#W z!nW7OKRocg9%c{?jp$$cyF>cZhO2 z?MaZ34!A+o<=8HC=OJ>V_xnQUt*d{xCQ?6$o}BsIskiqz@jNI*qEs46pXdOtDE&O| z-AP44!%@qip%D_&EZa;~vBUAEj<88Jnmpis&)#ZTfm9kSPVDL4L}QHd?{!IoN2!l7 ziek!mhhxJ@nA$GMc{akt<8?Gy(8&QpnYT}*b-aXvIsPRC6y>nIe%>m zwe}w!#ac;IaUCb~DzDa8dx~@!W8MHM)9>2XK!;fm;4WC7c+nZsyEEhW!63k6n%+gs zb(lxq#J9b_e>6mKtR9B$BU(Itw%7ZSQ-7v?nYOshw2zNZkXVHj>CG@Ph5HPr^H3yd zVdgi0fTx$aIVDB)V39!&Xz~k}E;L0?o1M%HOFMt{;7j+xmiPQEieBjR^e+{mC@O@oD&zVwFiTo8)OpFt#;*fz zPU;eI)4HRHMO(I0ojANT05 zo$f=SI%s6%T~E!0{|T)L-~0JqbXt%S%7%GMMl?|HbC92!D7M+5d$#8%?*xiD6a);y z+{bz|Yc@j55n^29UElfKU@%AJeIOECuC}hQ6AcF*d;O#R1Q2?>8P-Mnn5Mb5@U}bw z1zZuvBLq76fT)vhcY%ufBt1o!pSk1R@1>*ql8uhcKbd;tu$PoNF9?D3XZ{Y%U%_(t zkOs?}^QoczyYTkHh4BT=P@HROa&oe?m+gS8BU}95-ri}iW)o>T0jw$;sZ@0^uwEBh zj2pZcoC`t&I-!O_RS$`my8%t2Jo9vb<`@XcV`NdzeDh zehx^_A3{%8=lk3ixz)}H)cn=8mPmxyjXp;CTc$v&e0+IBZ&u;cM9Hps7BsLpfFmKE zGw;=6DC2-psj_)JHbdn3-LX5){udPcru`*_q;HX#?p7u^TJRNFqf7TMvewX01v}bLaoZeYrBXu$c=s{v|E%QS)l=gxE z2W$j1e%JVY@7`|=K;|trUGG(IYPCm?!}y8jw2YzFH*efHKHMJ@fnB%*0oBBR z=rR3@O4>Y*p0~fN{EY&0Cci}fpm)3koYv?vSoaV9{z+4bV1Jw0Qc9ir}EH{ z1m<^?8sy-`3n3F}e5P&=c8^MS{K17)k*nR}4@FL8=Vq${n7{Gu%qi;^4jMhrI*NKl zK~lRD3(}Jfk%fxi^hKjc?Ayl~w)d|HFdlm)W?G$j z14(S7O7R6D8yCLR42m`bocf>=f&(Ty#c9sa5mZwO)P9tO?k(uD%f&7&)8e^ z%e&ed3ZekImrbY))K3zwq(aljx|^E`Zy}NUZ${lsi{to<#_p@e9uD1lQLq03mBV?- z1agRTS^Qm{G(O8&$^{tOxLMTe>8f4Ak4~P!#O3sOOv1~2pHa}XP&Jw#$q|0&vsg8n z5+RR}uFDhT>f%5CCtH_*vWP}S0%uz5HrmG~^8?>-j_oDqKxN79?41LI@MUM*PC4{I zipWY^6wQo%8U&o{eHPJb*-XZBSDwQND45TQNuzb#BLE^`AKaG}uzJ?=EBnG-HZBX> zJ4?x2YR&Um6meHEk%RGe%MX7NN3-AfH1XoPjwH1KBZ3NcC2rTYbvfHTwI##A*rpaZ zf+O}b>?^r`)SiytUN{qFjLyxy1zFL<>yZN<#d2fxXf?R#E_%%!V%2D}8hI(2u5x&# zf6fiWVc&cSo?U1F-tBA%St{-41h0SU>go~}KuHIX#L&I3io5W`>=OT6IZ|!u#Ro$5 z^b#LPX%LD-8eNc`4#;a?f?5qkK`5{coyapO3(R-$9Pq6`*tsRhJaTqfUq_%Pc4q?% z_V3EGiO@uxpWZpn&|vhwY9-na0WTG4{9*g(gwZ!j2Y7` zN))8U$alpW@ytnzzcLlPJ+K&+SZXA`+fFvH;6Wjk4sgX6Z%#qKe3G^AdN)vgbWSms z-{?02$E0{oAf9z;xo+GxEghdRkt_szkDEm8yeP7ocgC)!szK_8{LQ;Y_!vwq||X z0P~L1QxBvz`26L|PDsq*@;oq~Lsaggl}!4`;AgfiL0{e_{N7%b1ff)b=K?PPqTv<~ z-@`qh!=~2I-yqhjdv1_$+&ca9QTioz*Pg zm$QR=ac`r4?Gvw9tFCPOZ7T}wxFT(+I#la_&}z3XW}nVa}*!zWn3N= zHZ6ftsI{)%r3{MQP|X3Tq!UBni5X6wfM&v7UWzE&QKHhkqNth}0K7b8XGg`oj}>MG z;m3{p3o-EE(5(UK06F@6NTH3&wx|5Z2tPjVdkJC>C#M%!8)#Ewx`6JmoPey#lP2!a zbq^YZ3KEJtKNs%+`3k*^fy@usYHS5PAT|TY#?v*8ZFj~%@0NG}tOJ`KxS)#=AEQq! z^m;)2hcVcTa9G1mEICaF<`Q=YLF}8VuUJNx)cq@UZcL7Y$@bxRF9b|vYo6c_M<{E* z?)aOO8*^2!>XCA-8iL^=lNXtdFpvo_r9;ofdIA6-Q z(B2Eyi(Z&%pBR3_+y>tzUFd$IlKpw>2<0*#vcur<-lf!A+8StTC)U`~f9z?FNK+so zEIZ)^e{E5gvF0qeYs}uT3-FdE?J=(NaQh_w+y8U>S-9Qp-`j@`F_Bs0|K2XeguBLf z>)+d-!ql<;&+Q~|yWIcWt_8Qh|Mzx(1{~%K_EOHRmg0cl+Xs?gb0*(hk%3KxcXRQq zCFdjZe6$PI%rJi)S1+jP5t+2B;@||MSlMs3mS2gA^Lx>rQQ2?plAe= z&5qgtACs!2y>iBwPWs0GxzTK ze;1V}O%G#5>p7@&pWnulAkg5H>P*xNeFyl% zyZy+uyK(PB`0?2|3qZu(Z>eh)UuJ|I9ZdT0P(oEsO#^R-jQk)!bmdw~(0aW>rif3p zhQ2;DebS}+{^gDLG2gp6g8M0xBC2g(_laOVCw(e9za%M(1nb9{;f-afWXBHOH+b-( zTCb1iT=Qh&`{cu4(_X&q9~8D8upR^Dq$_i9$aed1@5fcoMj|vww)*dtPW5(KhwU#3 zn2cNdB_LIbA`p2xD5_vUSbFF6(aaK6D5Cm*uNsl;GnV0i~}PVEW=}ZCErEuv$9+!5On7magzBQia|XL zDf~S+bmQvpdfffM@yn9tqjpzW{NRV#V;E8?R2B4B1K#6Vvi-{n6Z(=Sb zFBUrBham#k2vIa~+n?iLsK%h>xl1M|mlvYCnaQ_V`vE$DsqWkbFOa6f!MAX@zeOT| zjdKLq1h$=mNDm>x=J_?1q>lGQ z2>t{Bf725)S26fC->9F*gHUZd&{~%Va>9@ISGtop>qm$py;}GAs6D_fpyGUZd++i@swP&z~YHBfI~JIfVrPJ+u4s^~tl#PcBo% z`Q!sAyx&DfE_Jy6UbHPe9iqSXI2mkEE^K-jXK=oNpmvxGaPw0%i|up6ytjQ1g@3$; zwl!WKzBQKO{%3pspN_yMfnD(t`}aSvL-5>#KCK}}8XQ({?FVyZo%SFIW7>g)=~G@A z$cc3agW6+w4oFWNuO%5M_7YMr$%tnwhac;SwwRY0BH10mi3gTP(TXvzTfR(#4~5mJW}}KzME-l+p0rePit7xj{w$#zp~LFzj~# z7rmFiKz#qmF8t^Que{G^!fG2pUEnT=4Cjy!pv<$Kx$IkQyr6ws``6}dypUe?I^doO zP@QIS!id-|q^1`DY2|k2=j!K=X{Pl&JUsXe84wjjpTRdn9>tXi>Vf^PP& zZ*wx)HO%L&Fuv_nqtI}HK|#im%;Gtjr%t-~B;hoehS|n@3|H>{HF(FC8 zhnhvy_S~OC=n#GOh0M$E7CexM)Rl1k8LonU^ETBEEyTi<&Ul{O-=h#!Fo|Lx2+tov z9KnXtK4Ucr^g7=E`_pI%Y<|z#wZUmfgLfnf|G1H16Z6REc$3*BbKq(HT(*CFTjS3@ z;ks9*B8`jm>V=7FdGQxJ^=TH8rM@$LG;>NfhPcn`C~un@xLH5@>)Ua}8!*%M8J;&b zQ>z+CH!IaoDTmE~958JS*-!PPYgjGKSzL`GGBiwNm#yj=w+TdkynMXLg>{ik_1qPO zK5^7>iGhesqR>b$9SotJJAVeG=xVdw(0mPDL9bfg)Omjs`LzB51ixWLiWnA`T2uPt zgSBTl3(4mRRbx~+7A&3fh_s|fNRTU!uI!G)04|<`ep_d2F9&GN*2ACgOT{2r3pn45DqFIM2s2gTjoV1~QimjtXm1BXmMiDz@MljhMLItm+r z3igR4`y9Iv7N_~S?=wBm0N)9XSM_AXCN0)rQo7uC*Q^EQJ&?ZFLXTHtk04AopL)NJ zw$>xr6sTF%H2EI5Y0VLagv)Lf-*cIaJP{Gm)P0?u2sjAU7geC{8tRuEFOEv?oP_8o&L)-lluIofoV3 z=$uLoowx04bw4Fg2Wl%O=!EPDmwVGlv>ZxVGHyboL%rfvn`kWm3SAg)LRDx?OdE9J z{y(Qp%kR6t(|N{ct+gQHoY29LkP6t z^rXo&m1lDmXS-Ah+1f~v9KD=eSO;_CVH{y@>*K~aQh#gfnY|xhcp-j1=I+JsbJ5HU zNia|n$B{GN_yjYiViVP8`Kp6R!uRO(wzGP-PrgqK3FWD~k`F|3&sjxuR1~1B+QM2G zRUH%W{{XI~!z&zLrVa^KEOID3zD?EXRR-nqek{P3gs<(8mUYRxSBKL<3P@(Z3U$t; z;M4tLbM4T30tU;(%+Up&JV%G4=hfBDi!mL4tjA)`Sg!i` z!NNyNWDGd#)eL>xv!>%-(eRVJ?KF2I_pidBoP{J}Q6#x;Jt2-Xu45d3@W(|Kz^pNB zS~O$>7W5dByri%S?JPus)75M$^C}5pHFR)Xh1yQPcg3Z6CShi8H~rADQb~Cz=KS(s zr|Q<)a%5}NuMJ=Reek23Q_o?yyyR{++rzUT2`^TlWjw6M?vBJ{+3}g7iOcT|7+~GL zW{vjWa}OcpWkvzx6Wy+w2AN_iMV~~R z8k!m7J5NLRz)F4Q&b^MMxkkS-rW-}V-C#i~eTLU**|!o6F((sAFuWIwzptqyupL*MeigAaX&TtYNI7;#=bihu?{FE+Iv4+1@4@V=St!aD|?#tsAoeZl4$# zRilaVDZ+|I(`BDr1eZqYpkoTRM%8@8cB)Z>ELO(qP;M`|k&7C2P%luRIYnEJDmMT0 zN}`w zX%#{?3xXSz?_#EPJ4_^Q&5`yMZqvYQx;^wxXSIAvYNLfLBbDoXGWI;95-au&=IT(y;1`>?3t9;Oi$rW?$pwL9cRC++ z#qwC9sbll)k)0H52&ycx;p525FI$3khT9>mwK13CDDXI<>zyuU=(Cm4J0jOx%8TqS(hzZJ+hfhY9Es%E@o3lsmw*3S5#r{p z4^aU2%^mC)19IcR!>0FaI3%hF5egpth81$PhmH(>svYn^{v7|!nf>{axGL*LQwOks zHEeIHQk$h;*=r%zRpQl67o$HuSvi6t3}s@%frYm-k%D;V7Ad1L6(gEjMpFD0*oJxj zuQ~B9(DxMcbPkz z+_f68Jpt7@t6G=#3V+pBA(r9qjdKo?H<*hMqaGi14nd6lgo?jr2jkYHvp0>0m|74Su z@G1n$kJB|C#M9_Q0Fb1UeL-#NIL+v3w#+0ejPl;6lapd4tip3oyj^@=#@KXnd73o& zyHc5;t1myzPWf)m?s8YisMC7nBT*EXH23!K@}68g93Q-jX5Q5_YU`Ba;4G>kW(;;o;5cdZUvX z7-05Yc@9}N^l)Do%`NHat9=&O|F-hxjt*#IiA6m2>QRPvools|^^$gHnqYpp*5Xcp zv@i8^oqJi%J=FVYF|JAQrQuyozdIHZ*ky+mZ6Cud9$1pseE!Ak!k*cs^rMS`zNoXn zWiGY!C{~IA2TY9Z#c|e^1H2@d`0J8SvdWtGS57J7@u#k2TyalPZ3kbU(7Ur~YNJI! zT5R_@w-##L){~wOrjRUj3aJ#M-EiiYXudEa;Gu_7ZC5fVOemxbGdyrMq%A~NB#~zi zX?yWFpbM6;C5A^BE1{!r`;Xv4@6s^md?rVd5^cgMSS((){(yn+_%F+0P z*~J?$jt-*k5);nb=mgg25#h3$2E|j}#ppIL7WF6(kPt1KX%-d-KyHGrOc zT_)BMJj+ACf`b-xa_7w(+y;9c>TD;R?t8E2I-zUF_Ddv6kkQ2@ws|gA5ie|+wt!>0 z@N#M^$l*E+$7(@j+rZQ;{;&Z;Zh+sbq4j6ieC|Fn)sg5RMT;YaP>wCT3bjU9gFkSU zsOJbOT&m{%qPd(Dg|!hCDlfP3>Zdu)+rxTJ=J1)|E^${fqc`5$L3ZDjj6>vwx6z~r ztWTgNv7n?vgDU0rQQJ-_P8G95dZS&5wiQu=r*ohtAc?HnHWU3}O|R*m8Y!s{jQ4Vf;3|%N?ybdOI4hM{L8g_3f4C2oMH+ z?utt#?)K=Gy-g}*NbMo9s2hMTCkz5JAC1!9wFb%-3VVMzVb-c)uKVMW>H_ZO(U8AH*58kwL&bGiuugAH_-3#L!iGqMCG7VMZ&5?4I0J1fj zg5_Oz_ZmAcowQb|l#<^FLIKOD!|59o#Df!>T~_GuF3CW>_q7n{pR12MaW$2H$c6F8 zns;THQ317<%CmD{1=(xp=hJV_x3(ZxXyGVXt-vTfHCg8?IB#Lsp zAHJLmNyK(smiPp3Vd|2yOl5Mhx#kEIG<|9n%6)dpzEiFOQsWDi>%>+(TS*$siluXR zz=iW2is8q~OL~0iB-VB4yAle)zdUet)k`eCMU4?bI!@kbAg)^g?f^0Kjwo8H%#l2n zyw?T#+d6AL~>>+i3lqH=uV7}tq^cCm1j}kuc%e0IgyMNSKS+teHnhlvlil{+ zyrykop*m+1L2I_niceiH*|_$ZD(9>gb&|AcD&FcMA^job(#L^(#bAH+d8?!k>X(td z4}~)tPd%Y!K#x6i?occ({!jRZ60uS92Z~NH?M@3JvQ3YSBX+n1+Q5(&Lcy`XW%uRO z1-EoZ7qd1%YiLM!YwT8?N!VT4KaVpsyNO!a73fJgMgo6GWs@XrZ*V8 z0?A4vP#^N-EmRe3g??%}UjnvcIMFY~7V*@<`6=6`rjd4!k(lOrY{TYDtOR*L&zmqs zRWl6mxk%Dtx*@XKkR81HAV9s6thV*Una(U`aXJ^|y;~GP*i$Hj>*%q{(aw!#s4StS z_9CTORVWgM5BpFpIbR~W_cNHE!f_%i%U)7#VD#XmTDnwY0`px`t;tTVXMA?Rd=N5%bfu(P~8ys23R^&1XnLl?ZrgeGvT*Rj7FlUg!AUf>ey_AAW+9^k1glAIO z{HcenG|z*fNRV5m@~+uYwDoD3#jl1<>8ZgV)yIG7#-T_OOzs1>!RGqM3DWVwazg!U z!n|k1+I`;KIM6-g9>o}h?Jba;-B1Q1i@)SR00`y`={29hr$+pQdB67H)GxO2!748n z&KF{7?5*aPN##U~5;&D@kE^$CYCbrXA~iZ+(46AH8jT{M+VPjJg`;vcN=X-qtXkdd zL;aL+{Zz&#uLzr5M@s28!>EKaCw}93{ioiJ%ckl8#Y7b~Ass~#o27a9pKe8ojka~jS8))yzS%W}W}a>D!! z8%G?fcuyxgx$Uu9EHyp+UkgT5?2tVO$I4AQTXf$6Brl0iQM+>EIUk&{91HY|0w%>$ zk3FX9-CKu&XYSGZeC{s!x5g|NS|1lNB{p`Y-7U48?}Fm|g{9YQ66B;WD>-pK{Rb(6 zf1BZg$xpTk(#bKbmCnRpznC1IaBWz6VoGggSJ8TMu-zM=2HtuICv1zSG|uwaaZ<^g zC5K{I^D03Vf5wsL!?^LyLa#F?ZOVmT$mf!WKAU9Y9+l<69+?zt>S#nB+}NCHEO!7? zKq4|LsXdhm2Yl7%vJ1eCd-Q-e>@2s#fccQUhU2|2u~&CjN3=8u!ZRj-MaS{chDwQ< zN)nr%!@68w6c>=AmoXWRM}-&rtqg+NW4LI?{V|69=}#cSeb~Ut>(M&M8I2OPX)du zfodI6Dnw`u;yYrgivR+@~9uS;E|+ zoOQMjIHpVu#9Q0euk7BfZJ2pFeJx1W#qI%wQkg=&b5hmKS1MMKf=3Te*0 zEd~+xWjBsc&gC4f6Z&TLjN3z2zC0RhY8-s(Ro|BZ&#){F(+Yb7KkqW!p0W1Ut#bFHmpK3Xm zzkB;5sEi#7g1!nO9jRpa2~~0??v*7E#H!7;MH{}@ku8{L{)*v zw_Ia!*aBLAQ5oWXQ%Kh-MS}FXa_|)EvyQWbpBy}pg|Lqhz?|yY=67d(E!uWy>eNFD zvVjV?NteiC9;jVKP+^_kZ=6Mu_UqI_Z+S_Jz}qtY<~CPk9Os8Ekth>VDe@B^=hy8P zY}0521~bJ^-nj7hTW);wQQ8neO~C;G-Xgv)o34Ubf$Eo03WgbV3S!n&&R4wbcbCk} zF)Sk16L4hA@5Nw|M*OeUk#3tOz+|X6v$L_h_y2siMwK*x(enb?3`@h8{*K5of*skW z`2f?EMrNS^vt1vJDKmnCj5v(hhBS0F+3)XeeIzA94QtnDA&|lXSj%|W^>4eorwc3IfesE6 zsOUac1)6D;dG9F%XSTE#hev#~yz#Jz-pT3p+|Iqf{|5X3s|9Blv_FkDG z*(4*IV^i6CQ{s@lvZ7;CSxL5Im4qVWP$^N71s?(5>Fkk?N%LZ7EN_;5T_lzmPh{`aBQN5H9l8ADgDouPa+vscz~^DE0I z)KDUw`VoXI8C{~19|vPxZqECNkE0jk30{GI5p;w@bjvYHl2oe&d2!lfEhoHs-HfFf z@q!)a=DS~BoJiwVCi3IM!u}}@ep+K^4l-t}F6GHqPS4 z!@LX`s{av={XYyqOQC^^=*62^fJ>U@LB$9{2*ncMW*X*`Z{^8&eE80x7DZrhhn)!x zcB9W^D7<0!FlGQ28(cdRK>%<`U;pX6SrVKmvux9F`iI7!f@mR84|f1DLmhO>;4c)q zCw-syrFBAiiWR`8K76jZw{rhgJ(2 z4iZeHNRXN+<=NFD%w8%8ov^ymWxF*w0`J|^;Q_y~Mr;GC#GJxD^?c=ip& zwho)4z$|1ANO%4LKgO}0%o>nwg1d{B{Q`dP$5+S)>G^lGq#mk$8zeNU-eXb*)JTkg zZY348gF4^aAaiRixe=6PpRgbSIq!zT`A#-Vr2PZWXw(0`NN|bTjZaFeHNfivu|5If zAy|p0k&L1kKoUY7Rx(OSBcL+xUVuWS>GWmoMOJLEIzEN`6_C(?bhyB)LFM+gqwDS4 zaRyH+bbjH03EsdPQM17DdJM`3fkI3in9IR>*H3vn4gPaVC5%XGC?}nbf$~@ygF6pu zo*59>09OM*Wy71H1W;=Wz;h9}`oF&lJ~;BLy0q~94)E{xmIqEF|4xj^o(dwC6R<}r zs3~PrX!y|J4wwcfguL0APS|#X-YF#V@NGK1PRBqf%>lIWp;g8pQ8q4Aa7{{NScaw( zY?K4~_oyRx*AC!gubUWy^nvprmPId}=j~5;JQHc;Z-K`%2QdQKt_ND9s2^AQ-h>c^~wpuzd~`|CdhZTFkM*}MxC z9*CGe0-=BA``NbTvWT4(2(cQwbRN{YVqEYQUrD3_00@Cd8GggU7!T-}0hKK+o5otw zsc7tbbezHrh!GH8bJqzLoDhw;vUb83;ExOZ+HD!*07LDNuzKR;BlzotVJ{M+h62NE5u!pOAW}0RFu0!q%S>(aH3`q@KX9Gaf^HbhA_9LPrpg}HIfQAA zeW!-wVikn)AjPsj-_Z;>nlmB1)4v_4(UdO~SQGW02jKWU@iQwMT82-cBT_~j za*kop(#Sg`gR<%(EnFJUp^rm4h-xzh+`I9+mt~+&lQ~lVcR+{w`S|A^e5XyIOjfv< zr3KaV@1519oNBP(T(nvpt$kK@EOiy6HqjkuHhn?R`RuChO_970+>?Z28hC%h@3m;I zp*L@VUdX&BzHE-P%6s6~V-dG8X?Po6XIg$DwK56Q>8BLBW>cn3VZi(QO^B?q_=go( zMNo@Z)1o%T^?qa3kLhRDxF0Oj8r?hOxc4dFmEc4+Wg=qe#HU zQduQL^YEVgKVtqVgFegu>dlN`h;X|cM=5J3xQsuVCauL!mX-kqw$-iD?$-S=g2gpM=s_WA0bMxj11# zdE{vM99*Ly>qn$7u*-km2X(IV8&|#u&2vAnRA(qPHvzHNxbPR`Mo41Tr{?&A$c}rI z3iJMOCudUO63yTJB1WDw|8VQV4EVrDOE0kBV4IGDR3*6p{X}YZenmV3lq%N)C}KZd z`2$s)GsH6~talrl=?W30t8YL=+ih?kfeEIYolTQ0;A)f|g%A-4*ZclsuYrLTc-aMt z)^c+)t+!W4!qp;XXV@XwaH8QVl~M-D32OSd_kN8P0-8A8gMo*t;77mLVZEL0vJl}+ zn)P3b)&k?}mxv}6h@}}9Vz{|&*LY60f0mhkB_;9P-An!veTJZlmp~a{=@$Ec_U=y@wy>s`(Wv$avQ|5T->tH^ut$`M^HgB^> z{FC8wrUeGlQ*ZM*tJ6R`uoi$Zj7`;fQ~KG8%*N>rt%BSMU-?=C*&|5i1HPf6)u20w zU-d3rEHofY6#R%U4u=fCC#{S5APUPDp)hbpDxoTj^Y!VF(Lxo4vZ`YoEVVy1;8D9V z?TxL&RaddDXJ1A0s;<&2E%LZ5?zo<;j@v+VK|R*$EelkkUWxXvQw4tc2jSmpWZsLE zB&~4fs89*t3NsdfZl`7q{mxyd)Dw4ZS=|ypYq@duEMAPiep8A@oAWdd>t54LBXJq3 z*fYDJpc?VNL4C(&?rTb249%ve$FCiXlQo zH|te6K+|VDF23bw#@ag5&D7YL5;cl^-R@vI>xs%bOYf<>j$b_c^Fe{HbZ1=3ld*`8 zxQ&s=Y=;&~IyT@ie#o5b#+Jc~I|0%M!j4x5pitjL;&-AbrjIe3?oAJX2CV198tM_I zoNSe6?Q>AFQ^|=H)nwm?SEC89;D?3k0l^L?2VJ;}E)EW1gBr!-6dbAaV-+F#%X&Rc zo&EBiTKr5pZ~Ab$etNsVyO@k^R_#mUdQqo2~>077Efbn7zpZ8N# zUo-eF`7z#2_J6+2q{2RtC!1u{>EkaWTc}@52lh(vo%*)=Y#vUfi{uX4Zk3>3bHC8@ z0`%vwb7&N_y4}UAy?Z=BQSMGiXu({uuiu1fXDls&*s{OxRmw$YS4{vQq8_ig8=ww}rrii&(e!CB~+5|YkEC8Zfo>Co(Lt1QV7a^V`O zssNw$tce?uyAg^|P2#(Y$?xSKhtN9)vl|v|Qz9}Cs3Me9?RZAZAD?DJ_sxRv%k$y+ zjHRdd9RdxyznDsy<`zdEdMfiXgeTYNtr`ZSxHo>q2bRr5Eq1`Z2{(%MO^nP|4SdQb zqT+8Tqr!-@I!k?B@DF9<-%CZse86cTo6YT{FPepKL6qwQld%e4UYPnU192#)3o+SY zR82uA{#V|UA9xkU$1VMUt6v>co?1G*EJg&fK&BHt5;0uIk!;Sae>lXPr{i<4EF+9O z7$Pdx8j&m`2Y$0hb6=Pp-*Jn|8NjMsuSi+OwvBTJl-z}<(o4-1>^!Zt74IgDEch_0 z#ufbe^Ux<}XNZ_T0OIECAoFpSIe_RcCaVoO07>?m{mda$uYhEPLOFKg_R>@6RI&m# zGCGX};X%Whoq~w>*9h7RBR#;TPcWn`%Y4oFA#E2%g+ww6G{ZT~C>?ziM)4i4vcblG zTZlP?BTwhp!|a%LV9!Gbwc+OkJ%i^`LrGAdx@AMz0xc;$7m>W@E!T|qeLUtG0%XsA zFKBq955cyXUHdUQA6x^43Vd_+Py!jUqL;-!l(VdTrcT$!?;O8KIM{<5T}Z}e$i0je zJ7tnpdUb*I3#!)dux+V_pb^5C(kJrl3+m3_^sbfDg4~o%JXFGv69nrk$HqaMCi^}U z>=SRpR~>nodtx7FSNwq%S9;fAKJ^=(Zq+$~J}9v^jfaJ{jEAM*&2L*C|7F;hvD6jy znDe?7K(o_+B}`cic+B0g8Ebrjyq_fBM$=v9qVJDWo55+pUP26y2LHY8>yvx;GDW%L z*dfJvVG|B<1AT=TXgg>9y^}OAq2We-s`>~Cz3U2OH}3Xx@zNzZ;}YR#x->tww+5~geljWgq#7lilg-njyJ~4EP`(;Jn zVUG_npxk3Tl`1Y43GpemXZM7O;ynW6Kd4CmP&Kx@K3Wwb40K$vy{J5HD?0-#H%B(- z%#6_XkFTdbT$cRm{;4-xBt3)T%ITQoGlZ}1R0eJ2jcLIVDy_|7aQ*k)qc@d?&z2s` zcX}M4Sip2f&^D+~1^F%0p`5h3_xExvmrP`iV)I^4Vd%qjmLlL0B+Pjz?$zA1A*+_bSK z+sUdp7vl-#i1(KltzZu_TOqZO%X~qrp`Wh-^)tA7Rwq7g=tyKg-)OtIVi>e@nb6>! zyc;(3z%W^-^HKjki2M03%HQUcr``2B%uR3sg*MtnUJti&{tBj`w-6Vp^xn~N_*_|S(BOQN$bo=Awz z26duSd2n$$EFFS>X>gwZ2q14BpMpkoZe0;0DjXcfQ^sg?42jSjR$qfqNA5@pii0*M z(6?~x`GMY)D5H^%eoOML0h>vniKGeM}i5*d8(c^{cT&6~21=5X6(H_b>Z_P#u()c^6) zT7!X>Sux^=&(~Aogrru3hL-0oWY_o9zN&*K;I6lH-p42p={%yb4O?5ooD)7gfdFji z_W;2chgpjlQ3aO^^X@t5`C1I*P!%`m*vD%p2tpey6awn`w)GVk(MDRy-L>f9!uyw8 zZ#-<39V0&WOMkp6uUB(z%r=}4MFS16NA7xC!OlLp6O^*561aHV zuRXsE@j(mS8cAM%XrmM;WQ|9fj4)fbF!Qm4G~c%EHg7R5zkZF`wuIH{@iMe{9@M@M zhf*N@%PgHL@s{e1DCQjs7MJ){gSgr2u_K+$eN6UY$0qyQRAV;_8E~veaV?OOKPth& zqbpWd%YAXvLS+>U2ri)c_bd|e=_m*OmBanwRJk5qr?)^MYpjVY&Ud__lE*_*R#$@b zV`r!z_^QhLioSdc%==~NiIR8d2(6W5@dL>-q-jgp%Y|Sv$6NePHQfU1$u~@N!I#iG z>cY(2Pw#3;5n_W?yoHWP9aJ@+%siER%i8(!hm(l(LbnAs^0a3X6)P zXqHbuB9wYNU!Gr5?6<`~(g*N;vX82-R1;MzG2CsE*O4;We(FWczi3hQ6^|azUD`K3 zOBpWe7I(36+|A*asm>ZIs8X-$P{zjm-)FUMmR$yw<;08G(|TzFhWW!%Y)vMw?rRWx z{iXb>NUFVk!o~NGzofmVb|a!QJpAk(kmw9{R9$O(r1!SuMt%EUbHKcAgfnH*LUg_T z-^8bFk}UO+d%pWjllSOQeVbY{ zP%4ay2(LX^$rqJX<^J9B;duOWK!&=_bEU6dfR-ct+dSJD;?cKE z5dGV~s!6MJgp=mc`?r4nUOzOaAwLM(eCBU)nCIEnXyO%%PKYV*aPB|HjMIIP*=GSs z4~9RaJ$5!1>~dR~b{`>?{+pOnimTf0a|l&Y8+u5^@rq|CJqs zw_`NHwsv}}T*ob5j|o5u?#BS$SE&!I>xF;-+G4W!DN6(?PI(m&J?64vUiHz}5;Y?N zgu_N};W|QFm+}gx+5qC3OI2$%6+f1!13X*wYBsR@E}VV#w6l zn~B4yhi&4)!q1$x$W%vSs`UYitAVHs!`;yITZR5&wrfj!S`_{I!tv1>+j6~0oNi_| zH;opYk)^fZ+dl#a*Vra+wygJ++<9Sm-f?$J-)asSbtVPOb!mt(20l>&XYj-GaiP+e zyU1lVAwUJgq^~^t&G803_WsrjwW}Ypf%$O2qMlqRR#8l_x2~CQP4T~gJx=QmIy(p^ z;EpMA_(S2~a5yGS8wM;R4u@07f-_{GMTJ=SSo5|pB^fBGF=?}hno$E|ziv5?_HR9! znHUgY$|hajg2Y1$iT&aTQlJ=&kgk{d0e9V0M-3OMI2cI<@wP%M`efL_-BUI<_h~vd~gzjv`jpk zBMAtj(+)0lK*y;IfuJH@7+Nd7-Luv^Gj#gZ}Py5n0qK!otr)$lXXFRL$;yB3({NWp)dm|1< zzn}H1l~*l^6RXY#?3ofI2sBOnUX{QMWuKIjsuR3GRrf%tqL1WA#H_;P8l%%vSKDPr zm~J3U$u`g|Sf6&)<`?O%cW#J?ul(@&Dp!9Ch%eJ1e*;Bfowo$}*^>;|4)uLrqGS+( z{Lb-%q<4_vw76)_eT2JwGK=RS^`MK8(Y4DquH%pIBlbwYG7XrIHPdLFoZNxm1NGXmX3+MA-mR-wWDaK3k9!0pXm&I=n0P z!3fh!^hZqV85i}<;HUZcCk}ZEuQXmxdkR@Rf>(TuS%q~IR~AZz1b74-6BA~2qkXx z`wr=@vHc^G6@V8jFso)=l0gZ93pOC#F-deiW?5~a!QR|%nuI#h5z)H*;t3QpZeRjheR}3VJB#6JpWIp&qUe5C8SuS96&+L}797~q=J;khE@!s$= zz%7#N`|5-CuX>`uHN5@P|82biyp6`c64ade7nQYXXJ)>~X%GbqQK8t#$I=9J#=Bh> zr~3!Gq*vPy13Vpy;J9HCF(ZAI3(pHsyD|sU60)}2U0y%XOBW-bI_@SVh8#be+=Iyu zkRPzr((Slku7~-k!%xK6{hAkr&>AHU~J8 z;Sv<_sV{;<5R9h%yUK#_KFIiq@CIY)dk~j9G=8vApTn&_*_2MIL2eDV67nwTNjs?i z-~BP6>Lbt^pYtJ3S3mt%76>F)H5iJ!!#@5K3XAlF*e0Db2VhH&aBzD)`HLm$3S9fb z0>H-(o<548I3heYnazsXS+P|8bj5dyprNc7eMI09B5&tMaer65zTp52=!G2dgPb0S z38SZ{-vk0{`TLtZ=#)P1-G+#u@0>}TRq`|tJ$s;?{tF0bC2hz@051q!ZV?XJNQ5@* zAI(e!|CyUSLN64+CbkJu>GDfC!lrX@I7Bp(fPDEu7BvNNp8vdth6eEE$?&v6-oHcW zehmldBxC|k{f2KvKTE&>iC<)9NC7srECbJCLJ~j;x-1=fJI%6D1`K3wRJFNVsVpz!{`AFh=hg1cf#DXP3oG=k7ek;U_*RpmQz{w_c> z9KM@PbSC8({7Ci{21BSSX$Jl|3~9U{{1Hi&D^Fm14{{xX0vvD`(nEP;skQ-V1nqWd zQhxCI#J&E{p>DGLP;u~>nJpsvR3;q(Z?qCG@6B{%T#{FI%W$1z!ET)3BGl;-!3y*P zr1`pCpm40vBmWnZjM2NSo5p2uPZRc&sActVFG5`*7pcEOW4j@t1H}#LL6?1Pq4jDH zAk0GJ2HOQtI!|>VDhA3=3jhHg3EH(W5$>2KC}l$Z^0GM#{O9yHPJex5{M$ zK>^$3cLAzPN#jy&gQsktp7vTLYF`TM=D`b4#{(fre>ezabeAAA>42C8cA|I;E>YCx!&D$$Jv03}L8P0B$sbWgy5 zv04&1|1#Z#Q{qG<%9v+wZu1r<4n#)yFYzf51a6*wtE$GX3CTkS2H&89 zJpU%CN`e@>c{!yUW%zzkfSAc&ZXJOq8#ady=>%)cfA|bm9p1`C zQl%|vNV5Z-EPw{K=~AuG@Jug3;0wGLGC%YoY~#i{BE+R_E8GQUo0xS8(jy*bfPCEZ z`#;dTfRdX36pV85QGO|HfolgD2qbR6YDy`(%^|k}*xayrE|*$VZW^oJTih5cA4nNy zGoULm%9BPv&xZ`hp*mS``Xm&=A%91Uc+m0PMF`H8mcBwLsV+0^6xBi$z+XsKKuE}m zcQDHAcG>|5MwlDm%sWa6nw7NzX_xxM3M5#1?Ze^fuLy~UNhH5Pq7Jbbb?@xL_c!@y z-3KcSkhxW_R!;JyQvQkm%@%;fO~@9%_!SJL2z*6Gxl+e{B%)U#ojZaO?QkFLL3^v8 z{a~3Rg_TFzi8@sDADF|iEo2k}g}TB|c)|Vxc|ggo@!eG?sYH>qE2$6)>7VP;;~!r%jX`(Z&<^#HZ&uKbytJ)tjvJ3xs!cI|Dz zTv7%L!OiP(g~kp7hPKUlF^ufP$EWeY?YfB=OueWgX0$+!W&0wmerUnYO>npT%XbB~ z!n1XYz&=llZzeH0KXa$?4LTxi818HehQSxQ;b|{M=9x`|5RWIC?lbwnws2Dz@>6u} z-nTK3-6d)2k0cZ@R>rDy8PB-C53%KJP~!JSbM!1TVjs(Jt4DL;s+yLK+sq(fLnG7X zsmpOuW4Uv|u?go+QbbA&vz8OI+1}j7DOL*_uLZEqcKyHceFI1GdHY(0c zH(W;OD9ZUbZ>w5LDJwb%pT5kE|I}iCNemowstNQ547jv>%!_5U7~iuQ;bO#1nc=O( zgY)vtR_URw6U{!i#_l^sP|%1b;6IgZFI^I~uDkTA8xA%5$J^87vUW9a=h%LhdIUzr zrro3MA!89MpNfG^6W(D;LnIUC`*$95a7Ejvn`+$EO&QoI%kqKXp60udxOP|b+`eLJ ztbxm2Bre>rpT-_~OW}CS;y`hTIyMSS7hDHx`Z>)H^mWV=!(cPj`ood|`KR^_8ZvON z%z)|3@L~zfNX(a~@a(**0|6a6+jrthuld9`yHlmntecl4xF!Gu@B8V-z;I#&>hN=kL=Ry=FTB5O;;wIfoKrg*6AfeR~ot}Qp*C{s= zz}5D+U_N4&JaT76C#J56TsGe;`{l`Vm#gg0fISGdUp^CDpZS*q_;@X$maA-|cycyH znEcL8>Qe9h^hjpOXOkyFjKKWiXlN&ssqw&6T7D1T8&xR{Ir@Cj@1hu5;b0n_MDAQ| zZ%74oa2FDQJ-cE>xIeHod_ExW3JvTGYx7tLb5PppbD0dL?AE$)1SYYq@SU~syXO3n zv-uO!+`N$S4cEST=mE3R3=+-vFG&{agc5nr&viw`N&+S=5bv=!?w30w*-2BzQC*91 zei4>A7y}7RUJBlfI0ez8Q{PgHy`(|quRHqa`|+sk%RYw8ACiNtuV2n>eYj4Brk*Z_ zlipLr^mUEX#yw2Au4V#z;1k_E>8mzTwMlo^h4Hd za;ky+sDOX-uLrV9p~N8CnK023%YH5A_qP6gXB>15SA8dkOe0Q%k6t-MsE$a=!EZDe zXUFmzXFK@22oEjC2VZX&Tw;UC?^N?nBVx{`m#B$x$fYziM=L2E{vQ)|t5vRvQr2Sz zvKl%g&2iSIC@$&*e0l5?9@DsoreuCyZ@L&K>p6(+jFy^!TG~v!j(3XuWj@3fm?Zjp zq|i-z66%0-O#J@Hs_ANm(2G+GEP=@GP_WD8N~nt4_K(=e!$+DxrvL>-;2wqHJm4Dl zZm6?hZN7t4%~RtwNCsx6ps%=8a1$=csse!Cli%>26P8-0D&?-cRdwf+5(VO}4F9Q7 zp)txQGkYs>yv&>+NEBAzwqknC-sC;b__^pFv!h^>z%z_-zHG{(W7Uk^b4{m!ZOKM4 zTJ|}yzew`gLrj2-Oev2<&ggk>_S!NP5ToCyk~q+ zveuo(CJGxHaGcH6xL&l-#Q&Y}G2%r=t2tC+8-1tqv!+{v$Q(~ocF|^5t;*rYqp3m%Vys()7B|A)V`g{9LcCNW{tLTWi7zcl_k+-3UyeAi>{ zCFm!S2Qx1KmTi%amu|cMqnL+V4M7jguVKZS>N|d35lS4ls4^ZK|7W75Q52rsd)26Z zCg3^9*8B3(?iUjWcI*r947Py+l9#hvTcx2l#Q*e9@$$v8&OgHzp^0c;G z<(FRJJoxZ;Ws4^7t_U@w6VRfJ1EVM0vzEZYeONElV?08Cxf zg+S`?+P{Y=Q>u!+m=GaM93zp=F)L#=^T5yM%r0q8xDGXpR@DsmWWWr^2bu0}g4N__ zTB)w{ufO-CB(p44a1Qyvw8AAF<`6uYYMopGWxfU^x#-A;kFiN?aT%*sT>j%tdK4#U z@`i3dPx;QB(!`ORr7O5ZSsUk74RdX|_K7eBxi6B>{ygB5XGuPrM|sU;xh#qa+W>{zl?kKD zD*F~mpGw+azYbuI;L^kzv$g-rWN#f zFxB`_sQ8RyqZCtKU+&g?BQQ9pCTU%oN=Vdem1W`HGr$L{6O%-VReMDHX0>ssP04)r zRn)u0;vvYy^bP=E#5>iVCSmgha9!zDY^c@DA$rmr@zSm;`KIhO;i|Wd!m)yuoFD9e z8Qp2�uCyJHChVDHy@hb?Xgi5$Bn0@II4LCbf2Q7hcaR)u6GBg-$JjisZ5}%n!g8 zdSOE7seQRvna;STOm!*4)RF({T=bkYPBt5fU-nY?aIEJ&NXaS^uLZ9Rr<^ZA2CuPY zLCJ3R0mct62MB_I22Tp$#S(1L$-Q8h4~D7tc6aav@hh8jEe$+O+o8=MP_t#n_dmUn zlJS7g)vte01pk^SDDET72BUstY$_6nx)eBueD( zmGM2%dmPI^b;NQ$gF1bcfwGix8*D5h9gwso9FLy(b6?O0J;8Ex6ZK7-TP)&e|J)9j ziu18sdY2T9)&A~Zq5SCM6HjVgNR2r6PvqF{VT@C{YuTC~hp-b!vAw zgbSWxBC%IER4O$JDp%q#FjcFcSpjGW4+AaD={Qihl`5$%k_Y5F!a7Td&`c8;D&UPv=o_mJ*G5##+yoIwS z3%@cvcw`9ofcR?|3)04(ODz;tK+?B)CAibu1!|nihnpBYw5P@tVZ-m6e>Mj~3KCSa z9WpKl+pOUqvPDKnP3?ABj~%;t%s^YiqMbtaPDML7gCgQJt5B){JY66{^gpF5F4UP# znL;+^LL1z6T-{q8`My-QN!_IY^>DL*-|5LT`Ut9|lZ%&9GB_0(>FF_%CJ%jqsQD_j zL}4RjR7JGcYvhfKbdlN?YtF$i!#-VZbu$UoZ#RjOwQ@Eu31!ws7jG zKoE`Eg(p^7l4l{9aSS9LFhc~#y(OvyI0yXr{F)ztMP#Ob63s!ZMn5&?Ijp4p2%_}M zlRIzn%J@oKbFIc09{ngyn{OlQlVGuvEo270y3@qN-Q9aE5sKmf9!12X+e1gAT=ow@ z{{2UfPmR2>x3C+>nNda_EJSp_cFW5i%~5UmJ^T=tXo(Mx)`Z3&l8_)I3$+_TmM}Mb zx`8ni=L@l(=($uePc4&jU78D>zpf%2+qj~*3I3Q4162E7FaY%TqhFFOD8pa)LZk7k zMpO@c05j~xsey@>dHwyLt@V#o49h?{nBafko3@-{lY*5Dm_hafxOgK9Q#f(D+zQaV z1E^wuUsf)P5$h1KorJ_FNt$w{I`O+*0fUF4$ zy9=tBWbhor!nKE*e?Hg*KGTG`T2WXZgE>~=0}uzhw3`%v76#hxyNsK*Nsjn~<4O{T z0lRCqqUy^=SdoFd8cdq`RO`YH1gVc#z<{*k&%mZAy#|L+;#87y*jnnBcjfn>%J3k$Z`VqNyelqhkHwAgra zM;re=)jXV7r;t7buU~IW*(mwB&!WaAa}Xb8mh6WzoRQoIHM5D983zK&uBK?t>EgeS zhUQDnNUJmJOae-nMFLNr2WR!!Ex4140o#6Sy5E7{N7LQMJsBTonJRT2L=<{0`;G)cE=Oj2(v8! z7?!%&r_(hF%wOVl@Q*UXU>*ulacqh5>0CW3;z<$wNSTQ_< zAs%RKY_6Zi&r8IoRE=VUXBy8dXJb~cA0KGq?YFxB(4_{|cTN2ckE)WJ-F1^hA^nV- zYTB_{TMJju>Yx8(RaHVcx8^m`_=v{?czT8l$cS>E4u{^I;V{pxmCvio1=H>B>cei2w>6GJ0G03x3|<>5BRb9+FE&7447NcJ0BRp=p{F;(sa{?DT}q)+{kH8m zjUvS;0z661b!_^K4}VK5$!))Ry8fSC)of z-C+^S**Tus8)*EJ9`~-#C}bL3Al}ba_)iHKTQx@rP@ar2n}J~x;Cn<^TovhWr& z()l1kVc?&|pc>?u8R8FmQr17LZj;mA$E2G$U9B%B(1!!GSc9*Bm}%Gj0U5BOgde>P zL;579J4estbkb48Hdil%=e9o^kJcIvKYC#^{x@WrhD)4)PXGnA@Fi+HICE zlIB##;VVWw;&Fv?b>v6xYpS6NgU+vHPLIjAf$$1k{Ae%>bcKbogzy1&Z9quGvYPo{ za24uKZI!VT1Ol(8LQK@%W;t<`q3Fkz^l~$6z2=u>jvtozOyeHCqzX;H|bYAQvNfTt5nYLzmvr& zZ`+m3&t&ouoV=1|6UZAchS#Dht!w>F1P3-|G>3#n!7yonrk1O>Lb#jQBtaIy;(5soABn$Fr!&JT4+90P=+hIc4nHkn>G)Ini z?kX7q_0+O`Pj)>&J0W8``HquCUiF8}#~%vRGPqN4k&_uRODz67Lg!a|dxHV!9yFSF zv*{;tQ(uurfJNcjAIFs zmBSB<`Rx=YQ@W2{a3+?%R}2D9G+Kzd?Y47=Uk96K0_hPy^BtyN%;ZyNg)N>Q0I|{z zrb*^x<=D?(dEpz_C<4@&A&=rA1vxpsvdJ{7$%DHYMbo3Dmk=5&c^PmB$Qadq7V`6~ zv$!SR(fvs}tNOW*zvo~=lD43W`f!QtCG**)15MDHY-!rS4L>55w$)?pz=YkE;yRX= z=N>HX^dk4N&kqe$s&PxJYx}V~k*)Btk4C=1fF7(dHN1!Dm}9o_7f#F~V9I+2;S;6N zvF~4EPFo+;UeM>H)%~3%CgCP`nV7z#gM~t{ zr}qCW4#wh?ZxkwRo$2*cQQ%H~!LkgczA4E~-M31+-^~)qOdgq%Q^{6abPd{{rp~ut zovAdRN5%?_(xT)BEHv zN{3mv9FAnI?Dx^McXo^AVqspWx}CV(na+zG$`W&ni2n)##olyYq|TRJ{q0ISxR~>?&pPDf9{i^FHv{PG&MF1+I8{BQjl+*z;P{nDXm>ZKu4HlKv5RH5mscof{$1zEKPTLbLukr5RBFYnL0Pm zDUo>6RrR%FhFP7jwuN4CKE%CCVFVI3L9_8b`Ro%YRJCu!U!V_pynbDtM=eTUx!&kh z*Ky$Go+R}^Q5$w_3vf7Kr58?};hkzcwS$u`X$TnMr9f*jJ(WNzp>Y zAACF+s~V2BG^yToq1N2b_uVjjS_ltCZql-S2B5;7g96d9)S|TEBwvt>VacG zeRCR4+-8@VigCqR?<0Qecw}fzs#cjc?so z+e?jGAF363@E&J=e8+yC73Kw=Gh+Y!zV-B1?aa!tWt$6`4SlumZRDKbXy+k*M`Q{lHZ_i#EzLv>q+nHX5L#!h(83_P zoNHR6#I_1J1#_~`y~ciwQ)wi{U0uaOmGB0i>10$hMD00HiMn{vl~CqJe8#Wfor6VL zI5PaIH3O)Z&3|e$RMP=O|jhCC@R^^+!#N?~cz`V3UNy zqR#Si)^}ohscva5=#bha%MzRuPG>Q$h;o_oG>QRiaZdBMbnAaY8ZK>of_@x%aCQZt zZixrOU;YR;{r~=AcplF2?flGjQrrI&KDo&ls>~A%%T;8rXdmAoTtxIV{L=hH%?e~c zZ|7w7`^{=&BpnDCp(e<4P~G@<6IR)tlrwaai1daJb^3o)M4;1g`oDsr|F?hS)*%2c z0EeDpM_2ZXMqJstwVwJQGw|79YdXf&ushvrE&_Fw6dDhfqex+ zZIIqn3(*+={j=wMxC8gXazLoXW2-xZdn+Rs({BaBEZ%gZW{}`T6wXk$#b*0~f2*En z$r>~RhUyjse^*r_e?ikS?e0c8wS7!_G9y0HxN;7Z|AYVgnt?-rCj$MO;uTJV-G)_(St5^Pq!5U}P9tvz?r*2NdkGTR=TrfaQ8rwEL!r2?$pv4S5x+ zZz_O=+t!)^cXCAI3}~&A#GTt;fTkLTEh67uaw8^(93@pYW=fZ*LSoID9n*L`@H zxfq-nz>F;yB+z}Kloh)O9QuwzzXFUrJP%KyjfdeU(%LS&0Ra2xJ;f(@;nGiI#)v2b zbT2V~^P+`;WrgQ(8(tO}o%dVg+|d8Nrd6HutH0o;z?&Hrs~iBw+B61`mg&e>NFtXI zE!?!i&+38_*9NRJJhSKE(aZ9uVB#QAAOu|T(no770Ox?>g&kBXfbyM(hXX&e3|}=r zktbGM8pL3HUeNy9f(D7p5{Ygcop+E7!xSn6(A!t&GpI^N!Ao?00*T7)?$g9rI(YLy zgnJQKupMODcvYlZmHYpGNohETL6!gv`^(|+$Wjb)7wFEHd(6Y2rV@kOfa&J&I+Z`m zF|S~%Tq1enD+b(dJUCzmqLF`(^x!Fo;;VqI*zBQ^kIH+4mM@gsAcfGqsQm|28+H~Z zCPW1Dhaf{h*#JFT14$En?g2Pu6g$9=51-8xNbw^F_ysQZ-K3W2%)c^ zf!hGSs{=5S?r2MXhVNfP=sB{>Ke@pM0t5Z~p36e}Eg-oeA7ILZ)otu(eGRZhs0Ly7 zUJF>lRG^9lNn$imkiNU~+#%Zh>|%BH3L)9X?)BvjMe^j_suA zGHT*QhH@BXzmRlZ$)7uK@#vH)9$PmqRwKOk(8Dyx4@= zE2eqK1yL7&iOpx6Nd+~8jxh&g@Pcpv6!Skl5F_ozyaqzki+vwRa@gPk?ENPA4HMlxql&C z$KG_92NJot2wxd{OY_$?xy$cvdrs#>qea8vW5FZ${4eDg;sVQ?Q+n5YZT%=gI2IW} zbJ_y;pk4BaQz*^oK^v0-zfh;+5U?{&C3@($Jasi|TNKM_f(BXI@ZQ^=Xu#L{Qz)W< zZ|^7bb`H%iBWn^+_&BhqjGW>J;H#oN^G)v$8x3o{@sd?gX2qRbO}XS>N$fCZdal`P zWoCejWPC^UHiAKp{`@ggq&Cs_L0!+$+M2zMR) z9)NvRcI6#VLhgg0YalUXV@lDs!3{eNvzL~o`&iOYBycRZx;7umj%pHYg;UW=po$r* z2B9S4rvw4Ie4xJ`mr5D$17MA`#3zJOW6{-}cHNd_bpF&{7g!jv z$Ch~SVsbPw1J=0yfCh*sFcIH!1^Fo4@?T&acEd7E|B?Q9py{O*4YsNy**C7H%PmBH zDUi#%%fz`Z^p});KG-eCoV9CAPGgEDg7Mt^lIpW{h(O34 z(z+>p=0Y??2|?G9b~(KSGs=ey|NYP&Ld?^CnB1&Ws`IFm?gr#o0Bk9DDd`I^v54g3 z9?3=#IB7=UV$`86fIt+k7!LOUyB6P~wjMad{Uw9PK-r$B5Wr68zmMCz12hxaTv(pYe9?H!yb|WB5A75kl-ye^6vY`VpUE@RPd62 zSQOnHS84B&&AW8eA4uUcQSdE;UCd9C`75e_?oGgjDK=sE;Qm<9>j#cgk(1B3fmC?= zH9hZ@fK^jf%z>_Y#ZIMG-yjG9%@w@tLbMXoL}B%9xj*CGPXt{DBF?UQKVVUwSssdJ zsXvredGYfW=CeFmB3xuo`nqnuD3i0)a_G=pJmA8zVi(d61^$4ITXt7V>%xRt2Z&KC zYr(Tome}b{bZ{;4;+4kncwOtfXoPZXa5JF%9M6eAsHlQHQ8hcYHY-ka|8*vD;0bzcmo9F-7{zhQEi^gi^Q%$Nmp*-{DSm zAOBs)p4lQ~?-fTj$KG4XO3FB7uMl$V$V~P)Hc1ObA(6;DW|Cx-j*RRG^}Nsh`#sn5 zA3WD}U-xxiclXUX-|uI<->>!dgYt0F54brH<@vEZz2e0T_n*5BJxmk6k2>e(KgM!h?|f_o>xmYvLwmSf)ypVv`V)g1 zc@A=yU|=!K!<(N7F5L1cg&|{B-C;%N9B&RajNY!S1%Z9t!ndt!FwWgm0S{5TI$^@N zKDsk$SVT3+^6H_;pNAb{#=($yP)gtT`*0i|Sa`L!s! zC7P8o;WKZhyDHFdn#hyQ{70(_KZWa=qjnqSM_C~7oMhYR%OE#X{pY4kg;WxH(Rf1x zswwI~>M{2D&V$R5B<^Q58!W=j3qdi1l|6!SB&xZ6EO->{Hnp zyjr}?mKsp%#JsR{H*G}uPCwN!IAZEYZo>I*L2mRWe9<|PbQsXwI?^mu1@)$mwT4-z zbnkGh`co)dl`b|fRW434(RQzSO`DpN60+e27hROOlm9-X4|qjS^9gzedxJeDcFvFC zh$tqs5ci+#tE&mEz%=;$uYEGdeBQZUCnYf4duLWJ+4G@7F5>b*%a2RGl4Gl5iEE>emyPBb(8-9uiPx0$7pK} z^66$_`{_jPAr7<{u1V&?)5u@As(XA;-enTmVpTuzCXmbwx!;_ZOxVfQhQMd-+bH)~ zO`Jh@Dde^K^u?KBdwUDS)X6ZUU{04AtYi1Y`XW@JsiB0bcCoRN(!p!OLs}d+b+{_| z{jHMOM0LqsXVc=Bkz%9rXe84Yp5R>w?}sgG*yTPij@J)$qZ+w{D`}GE>d@|c@e8(XCWg^Y6HCVE2=|94{|u*D+~|`hf{UHh~OPY=^iQF00i$07+J1|oe)opzDdhby&V$N>6Mt+9`7V> z<-XLyiigEMDySlAU{z>74V#=nX59)fH0U>1{RM%~Dkla-^;AkR#YokHyzR>V2Y@pA zR8_+$HGK8J=BG<@hZd1qTD@EQdcyBF!sIte=gH&V{LX=uDXmY-o#u_%^Jh99zGVJr zq4EH_Z!piK6*v(1u%bKb%^y|n;Lfug^Wx@LpS-E|Lr4Sx81iUUy@mH4s{8)@$?4Ol zfmulKbLr2%8^Zh$#aBT+ph8AdAXA^Vop2xZ3k+2Zag=XPB^p%B7Ox#1yj;%1A}vjaC=sPI4uA_a}{aq z98D0dHXs>5JJ16o{a$k_a(I!kgT8dl-VOI<264r z;OTUpn0|oCo`V=iy&(m^heLq!|%V?!?5WnzgoZu z^+;Uj7CY_bE?!*IpXBNXqJJ6Cc>)5Rs#UBCB*F0txt4AKCC6`Afj_WL1#5~^@|C

}u=PTGZFqx6;3E4nX5eNJVC8;*I-Z zrO@X#e^n)_8sk(}<`8@}fM9T6FxPvi`5lr?ke8yL_etxS0mDzou~V&c`Y-8mH&4y9 zNCC<<&>8o`)&GI%L?NC?-4Np=o&%P!9C5b913?`1;3E*aP4fsehJ3b$K3>O6#gBC! z-FuHMc33DZ%caK8=r?z4w1PwbjfB9~M zNeCZ%ct_X`&JRe=7=`XM>}%ZEZ2{cwGSY{mX2rAiJpRYT%X+OL%Y=_xSXTqIEf2@x zPMC}cxy4y}ygtRu?Qm^n@II?b=0SIC-oCP%R&ZzQHA)sbGc2j4ecuWRB@#IxEit8( zywr+nTttr)utlj7{Mj65O~J>0B;qHi+QnuU@)T5Gl7U0XAW)RAUwP|D9T@VO%XzUC z)ji|kf&_;JIECi@y3!G##&xF*3V*!{TyYai-a}VsCiEgK#ACm16ZQNYX|pVTiZ#DXe9(~3T&48SD8Gwopm380hX6fyGxfG#tTpB_qsx%Cz6 zEj?f_@%YOd(esDo$tL$P@eH$<4j4Qy6FFf~fZK7>}#^!zOvswjHA@~w?4iPHWd$Q~>E z8;=3gYa-jDxhM~N$K!AGi|MBW9L!7UtO+nP^AKk<&LC)vcKopbu>A2TO>-)>I_AC9 z*u6~X#Xn1ug9;lF?0uuMCQ>up{4v3g2icfKauG9AoM9yLtG@?xCG6kdN5Q$nUyWL& zl!d2S%`Ab$ybmrjDI(=2=-yPM?_RA9ws4;{L5?D5dsjPJK^(IxBlviJgRwvQK6gM_ zv!qEVaE}C&xYrfp0;vl=p6l=9mbcxb;VOV0tQrXgbF<`+#H{E?R*l)pXW>AmB_~Jq zGWYmG>DiL~;_yS@<3}f2A=k}_3v7z5Si^<BJ$qx%P}j<%}t+~jk70HA3U4xMipDpvjtP#GMGFJVUjS~p&))IE^| zVS`(OZE=3kOJ}PdlCj`HHg~wv>$yG+sd`l}nL$fz)O{9L@{FC&KqJf($U!b2AHIEW-5l)JTnvh0MWmXj#@g7J7`TCT z%pB?KC5q2Q!TNgtYOKcW%sK25Kw7i#kkBqRN1NyFEWV|*JArLipZ?cEvz9rXQD_)E zSrOm+nQIrI*MY&RMn!^mq?D{N?=38pc2^qW5P@dEmhKk{Kxe-FO{o~Lr)gDz2&y7YaTKMvS z!Ws(O= zeWJLZZDv;h@wL6Ce_Zm~?c%8oorbjJ@$rycxG>;65EoW3ZTh_%eHOH?cXx*wF{A+D z`Gx16&%YLVPRfCVoq-Wdu8Z7E0H@sTpb^p?IDsH77|^Z^2m@q95TG~8vnG{p^A~jFr6l$yS@yL7w zs=Es9fRnblkJ5|JWx@PXlIS$%{wE^v7$HuG{N|4VyWfvrjiA;WDOESS{}wrqhwdG) zjfV?DP)9hGE?anuWf<+1XzJ0nQ^n;Xq1X{hK7Xs5JWlNOtxJy`Y*PjN=q(=BP2c9l zDUjHGM6>}<-SRuLzUBL}7Fel4MA$b3f?+r$O2C8aKG~osBM-j#RclS;99jaaG4)en4Kfhgzn><+ zm%9pQDL-{YD*qx*eGB9aL5^pLA(FiEbRG(6#Z=0*`yXG&8xh>oL$e)lLX1k;<+e~# zQ&$SttG5ou+tIs5iWdTcJscQhgQFg|Ndh%q+-_Oa-3&#zFyY4?hQ&BXVc0}{kCvz0 zVfuD)jY;93Rj3#4}-b@?c~;_VBnu$dYca+$aW{B2OzdU<7ua=hPcBw=idV2 zdT=5eTsh==RVH|z?soMU`Io^A9-=h!;_VctvhKf-T3ou`B`yTJ;afbdg}$FO(i+G& zdIzdi_LtL@fKeieeexbAo06n+cg?IM`YpxHZcsG2aG691T{9&ddyz+sqPnlkw{FuE ziT6BIWt&9^J9;KS!_Izv9T}!Y;!_i1hs(w0KhLoECK09286SUzPy48oW*p4@N*8Ab zJQYGF!ScMJGp}$CiAH&V9y?|ikn+reIYXEr?lDsK*z4qA>qf15(0%3F+9$X=JV543 zW~47kO;(>r@iXjkJ*4W8pVT1EK0T4M>(wiLyR0XLwPs-&KK#OF)5vqL>8BW4`$F&Z zz$cJY{Cmz~2S!Y#Y|y^yXeplDmE3toVWucMIKnJ(}amDv9@qRcM-yu9uC*3MCpQq?;9*=PpSqi5|JUK-~+?{CM7&it~ue|5nmNBk32 zoe&9+Auq6m3&TKY)hhn@bxkh>Kw;NBtMcMC0?ioI4~)d*5;2b#&Nv-P~eDxk(X{`lec%~940dtqMpatfB~3Mss@3@U0~HtJ%ibL zX>yn8y2vBln4`Rud7L3H1({Buy~1bz2|(3_9BiF{2L``SW0M)e49KORvImGQg_2nm z=aK;8!q9{FW|dg@h~IF8R66(pxAU4{WF?pccB}5f-xV92#mSu);G?SC=lbU{^@xr6 z4WKn!DSfa8_9A@DmsgV}kSMB2kgfQCsa(q+{cts305=3@z2;fSR*2g~yq|^a zG~_iP*!{vkNj~2Ju%B?fKp&fM3S0;+U=(LNWGp!Ybd24z>m-Jp%<9?QP*Nu=)x&W@ zLQ2~IjYeY=H0FeK1^8>8*TT<>I6T?mUM3 zfDmF4`WBAeGID;7{~`>!Q%BT}sNWm_TDwXw`4Lv>LB>N!&C%2If=C8d2|FZ*;-6Dc zpccbo1AF0l)$q!i6lw5i+$%+B08my^wO;}i{`TW8{2~{_C~-K1D8U8sZf=(xf;E!j z2)|JeNE&Af_(WeD-P5^PeO$m*>qm#i!HgOXOK`J-=exdF$E)uZe5Gc{-BOj*1Zxf! z`fhULVx>K3J}3Nq!@zvO#POOilEJi8twAr52qdd|sqFD5gzrZ87P7dApHLjB*gSzW z7kZNbKL+&2Kz05%e!u2am7V~sx#CaH3>W9&9>amZO&NUDSMaC^i%C0zz<`zq{+tl8 zjvcbF*WbsFEj8ydMSkHn~RG^$92uhg$@Y%nr@ZIhFoNaC7azD+cqx{=5 zh{Zj%?F+=+CMFX5fJ~1q9#FAk9vt*qdQst*1RDULSqB!guM=`QR{ePv5X;$P@)%uz zRY-O4xpXx}5aI!D{o;}dfprdO*TfPNMD!DN%MwBZB)`n#oqqz=M!o>t_`DfT{}wzW z=;S8&V*$#`maq?jrxI?NTMuN5zIj`y|M5RqH9%1=O`atv6003RP6rx&@T3gC9?InF z1^7mgWCskLbt|LQuxo&EJ}6Ryww+J+p7}W4T?T<Ep0)+V@T2D5+bC95a7$YPJmqrW<~#vY|=ND z>UrS*_mgqq6o|GKu~&BmWPLgiUf$tLD9*paB3zY;0ALL{XrvoLIFFvQ@d7ge7i{Cc z7@Z8r6AG$K`1LceU|loiYq+V^58NlHd>VJ<(#FBetk494ldI31YANGg?9|GXvrGe=Cs5{_dWO^*g z695!f+`AOkGwo`}ia=R|s%mSvxSwA#$2o^gwQKMv=v)bW(Jh2?PbsJ#B90H96)*@X zhi+bLh%P(EW!6H`FI*0J@12fX_6zF3%@An)a~PN={zIrVqU&Rqv{Wx)%3O{BJ`_A3 z0G1lRl%OoM<0oiET<@4&MG|s1UUkFeL7->{2mv{pTEupn=afU%^Z`kc_h1LY-KM9H zSsaI>LN2T&5!=d*LYgThIJ13#<>fYO^E?;pXGraf20uT`z!lb#SQK9y>}Nk$J6g@| zSC8X;5vjyQ2f&nrHv(ss*Ga)mRs_gRc6jnNVC}m<6cMv8{}7btzEEJAi6W{shGS1t z@Qxv1icKa2<5hCZ3?Xls0Og%fG{?NIznc%Gx!UIn5-tf{TKE2JgLGQbNdDc|6JN4W zv49ec)rV5k8k^Ml|K4ycQ0#oljbJk6+TYi}z6DD98fNznO_0KaFDh>huuq)cm}d3Y zl2N3%j_RGZI+7K|5u`2;U+)=W-NC*1R~VUlIXVn3!n)6oD3G*uaF7;VCCYyHafWF7B$2gmaB(up!P=vQu?`$eiPnSw1@TAl zz$>$fDT08m*^+wa3R}-lhGQRqQm~z96@tfJ>ZjlfMEjk8f15*?#IN9_hb&JVcTdwD zz@OZLB$jI3&eUseE}OurE?|&easay%vd*Hm0L5L_xHV}~Obmz86jO+;={C9=%>izhfQA>AucFC|w`aTo zd<9!olY~s%2oed#@Z&KX?z%UuPPV=cJ z_mmA|q&oV7DaOk37L)o6h;~F7gs%Y?&PVfjN!LoxQ| zhtRqvz<}n&F8Z^_ON3XSM0oX~X9Iw*vYUV%v+_Z%Xdcqj=lxjFP5p4bHJRp|B9|Z3 ziD{5^0^LR=-#65wfr$4o=Hn%31M4x@z(KQXPs5Hh?>4x8>YY9fnIC8J*wpP8%xHUn zsWe1jbaF+)d9{eMqF%)mlw?)AKe>@-75umf{2=^CMKZ1WupJ-CjspItUl;)?4jg@m zRIzxUyIIE+5EfH44S0n=ck11)pav^t&G;XvCDB!XNgKGsMOawsbrux-zlcCY4A7Xu z7Ep=kytTR*K>SC?{%uJ^Ay%tuxdis{MnBjGv%i@|OIbd8G1u6Df)tfR1_&D+Qz|_t zSx>qssdc*1)L!pNLP*RU64xZvgW1SZ{T16gE6bWCZNvtC%0LE*~Y=};Ww@o-U zsV$Ugt|H0Um9?GFaLU>p0P^dh)vN+VOx$_6v(HDrvIYm_p={vkpRq3jSRtc?6Aj z)$9O(mkkBy6$9ep>L6+k9&fR@t;{O5uMiZ92KF9iLSA4hY0eBFIDnBWZ-B1wYyn@N z6;;tT5It21w|hX{vl2>rhHDVRMTrcL%zgx#4RHMwD&k%c(CDYH?|h+ZI)|oX|NV%I z2NK2zwOrISfT1Q@>=Gp?^4Lrq63+N@Q#A$qrD5#P=G>yZ3u*4$qHqbp6M%FI)E`1V z3JpId!SEr5T9p#urZjf^U4rnPEaxr~(lb=<{B7e}?<73r5!if}xWCs!wuIzLBWopfmP#07BzYFNxM)N;Frc z5j|R-?+zS`vV%g~dTAUM-Efqb@aj-%!?|fq^YS|goj`Myai%-svq5Tr6i&b>r_#xW zNR=L4a1TQx?=_9>RXnwI{pQ3~@f{7U^E#6yflcu!eVa}a%G_#Bq6GJ~Zl7r%;3BxU z*amJb(BOQ6n_ScJSBxJY4rGqywcxSvw4iOWO`7*k|D?o9i@5J5Rtoh z3rfP}*4H#L{lfcP)!StzAeCZ49ls;ypbZxcB!i)(@>}1gqHe8>#g_xvR<3hl*86jl z2V*?sWvIw%L{f~xKc-5Wz^KL6nQLC%asWL5smVQva9JoDAkxRf;RdrDH5X2)d zOgwNZ1PPkkDb;cvewbRUn*S#P1PcmcQiJ!jM8Pl0Q4gbEn7~x}{*LgwKUpk9zbcV% z)>hMfC7y9Ljk^N3vP^!p79EswA%8;27Id+1jy9B&3A3G8dNyr$@@mf@BG;xb%)I;H zFLyNwwL}W<5Ck)ZntDZ)R{&iWen*B`uDVtv=&Ve!3&WMVZU(&r5juofLOq7ED9fdd zq{G9}KE5s9q!^MktQex74ujuGG(yz~n>spo&+&yAIA6R4$++`KZjYgF0bT>T66}xj~krkg)Q9D0FUzD=*fZjpdxFH3_5=X-dhmSKoWNUdi%A?;t^@1%$7g_A z!i1js@DY$0&23)ACeobKRThUlIKvr0(6~*Fe*M9XD>07bf}|1e$B*CW0v&R?4l3U*@7wIr zZ~Wd$o(;hWUaRVJ@177;HHG!KbCeC&c`J*~yg7kR=lgap73{}ip?jn**}*i*Vz4Gh ze^l<7`8{vaZ9aF9!oO)G0nyN^P(>SF4Bth=<`ex-V8uMukhmU;a%v5T@36W8Q%qN9 z#Q$5EW`u=#Y6nZI1~<*mJUT!EoIE2Z_Zn5`@o=t~7l5{E#h)dVtp(&@cz;RO$Ww~I zYC)cRX~gzj=csR*>3*%Cjz!&uMzx-S9v)yLHzc&xRYp>B1EKY=Vz{ zfp$|FaKPaI;eNbc&;q@a-6JD5H9h%gA~>k0`Ms-HOod=QzRoslzmkiZ!hYy~!S&$_ z!+B!D5WMWc5EFwUIm#@0ODr@ZhJ~BAGOh8*V&a*7o2e&GX38MTDCc5rb4y#lO8^Nt z3E{4rWG~fEcCEbWNs+ffwJ1C2t5^S3YfBpo>2ha0%~R*=ih#~!d^!za)d_T351+v4 zl5cr^q+hLsdVb3{{yqwz0^~&9h&D|Bqqa)N-$6}={NPW{}bf% zLBj@NT|?69S``gx0Z^Tn_%_&8oI#|jc3*DZ>H%hWtm4$VCb|Xupkq8Dc1^V}P+W#2 zs=3h;b-tN@1aVL~oOe`SGr1iUeJ=*Ploh{a6bEY`_{8z@(UxjZKiE`&iaU+(+sbGV z|MW3bT9@uGAXh$XC71!3a&_nPQ)JoK62u#lyy7YIX(@z^|!>&v4(tOSA>6`k!b&00lla z*w^^{cVyY7H<8&x8SxxwIYSD6A=kJk6{+fOG7!QLO0h0|w; z0-cd911Ye58NlkIPv2vki;mFR@x4ME4wq{a#dWBA)34wG#FJ3KG@xRkC>}3#Q6LHO zf|4#iWenQdKh@NPk;D&=K*$Y$?7^i2r4&?x5Rax2ItM0{6@g*+yhvAgz( zamEf3;5g2h#=UZR2x2Qkp(IU(AUqDRHAb5V$X^gXoIWG5>%9=%DC z!iC6HGD8zz%5oVta6P~tk3SayHy-ruZ{sUQijx5n+Kd1M@|g6P`LbSMu^cp)E|I@$ z6Uh(+StT6G<&yx&l#W`$9;CA;q+<<*0MdXCYNoRaydBDL zBglTkZ^e)fdVn>7=j9Z|B`nYHKKW_OzDRW*XnF~UW)6h>hvajfD@drsFxs{T2ngNA zz;x@>iJq~}lOCf??r}ofs~UfJy%Qk(I{hO-oa7GGX4{%Pk(D}pZHPXj9x#dnSIANi z>a4!>W6q)B=afiKBKDi^enjAk$+&(8h9XK#Nhj8WvFH`FYtYlv#r*)o=e#?lN`cq*(x8q~lP8ggF_^mkE^?JD9o&$b`{>Vc^ zkZ-@l!zSaRqr=k9#sJ#2bShk75Q+cz8g=V8x)b7IUX;ixG3QQ@xKIzdbX~hW1m=5rT5Ksd{(D4x*NKNb3 zp)_9l>WMcidqn_WE<3V@t!fub!piC>^7({thK1mlW7T|6WNuK0S_c_D5RG4rsjnW= zi`gGFLp$xiWiIpEV_XytV8ZXl4}+G4=klu59?T!XNC(JeX!u~SXlD9nK(7Rt={lh9 z%_7K_G=MD@5eYIQBwm0j!F`brhB2Y{FM={kj8EUnr~0=LovA)8t3-+q#XTui3Z~?2 z{gd;(Dbqsuw(wbS5cTjZT&^PN`Ze21ldpiG ziuQPMv(5JZ{%62a@My{cW)>xO0R%a|`MCZw$bG_m`BzRCO$nAQ; zXR|Qf019CgnrCR?SL*)v$FRYV8S?`t28!|OU(Nsiw}r$1Pxh@2AO~Z?mT?mUo4(X# z8}17a3#nwtE*~QZ#EfWQqVN`!h!b@75Ci^%Or#~9nfSre>>NnWH@~o7iKC49L5NLF zo`O1Ruf>m$Mt0BuC(0_lCO@RZKJEzexY%83wH=b6pAk!gES3-2H^Z9+<}n; z`h6fehRe_AtS*={CO5%zdQK3qW9x1wZU9{?w3Of#l6C!pra8S6=pdEwU44VUOb&HW z)xS?>rt(;PpIU$EH-tMv6BGrElg9|;^T)G`k&6NvCTWgEK?iAJz1W-n2E6U6@npx~ zX0e_IYN)E0Lh~#Lxk;EC;2q)p7f^5vZ`s#I?UC~^Kp1pYClbUmBT*%H43V`RiJVZmgnV_f4c{?ls#(rbBxP+_>P z_VvT+g)uWhi|;5@yYhkIGXW)T0R}M+HihQeXn>zq?y8`Om5!&k-|P>oUQP7;w{-tA zy``u{cG_oMQN;ib{$fT;ruaUK*GgG@L4jhoEFYK6=(ig2zpQoSd=a478nhO4_`qS6 z$-mrzPvuSe;fps7?QYATMpX9QeARxbx2-i#4a$N*((YF?fV>aHEVjYw;;y{77L720 znc9)M7oafot~Jd?v5Hy?Um%QKY=FiHw%^{gd1M5n)9*_L&_e&0#V7(C_rJft)oDgV z5UKl?h@EdA8lAgI`NH_+0^Sjn(C+WKV0aX89?y9wKj-hNoB*EL6Gn&^ z4bhecIKEpDBp9zbH5_13@b*xY*C1{FNt4Vcy|_WwTR@>9^pqG^1F zA$C%s#taLDk)iT1VJtfw6@;0D{K^1xKt;%@zo&&gCv)rC)rE3b3 zTY~BCu0_roc`jVdALf7;@aX;S0(X+6nZ71P(Dm!(@cVSb2cS-D;H2lG3}V>==a%7n zORNZvUz`aqdm2fg3xZ~!x##UdGx^KK215$@i+{&U|MK504PB&ARuT#%B@Y74N@fdO z86j~4Fk4UHHgu}@Idku8Ecb=QGvFbIX|xPQ7C5TI%5w*dX6}0MHHvk_f%}%pL6~3! zoy$C`OVxBq7oIayCUa6iLN$H*+c> zkR1;WHuuoPgDFlbCe4g=xxhP{6DL;as3tT!()4A4m# z57PZI-uSN2<3UX&C?s?S%9S^JfL6g+mQmxi?_dA@mCyR{Y^YTIt6iw1R@9#rAr4C* zEK4G`4T@%f!(yahK`TZ*!4Lf80A+eRvm_)dJMJfVJzJvQ|MgWJxCmfJ=Cwgv@H}MQ zPC1@R#vS(it8eFNLjkbu^PJB5H4p(EESq9LkUvw}{V&P9%pwwLWgv4bv!* zy4rTu7W1y9CWEMh3Q=y^vlRiZ!R?ykxFSc`8n~{4KCutec*F^Kwig-t=H_QRUY1ZK z*MGZ;YnOUfdwMNN-v!=j3&*y7ZGFY=G1#5}kqxkB&kBVG*P?(Ty8FH3Cs;m0C83Da z6zl73lugzmXeKBuTzgx&wFF6|9Fte&KGbH4V|1(Jq5D^ia(IHTw_Z2^PI>&zoqruZ z-U(tpHGUfTNGcw~+&T6q?fZ77cVgA;H88_l^TXM7BOn0H%vh7cVZnfy#pRvH93M{T z4B^3h>a52L}Z=A8VxgR%xqFH+fiz3AAjzKi52Y`p)5iQ&CgZC){SA zrH>{^z4%j0*3|!Qo%tf2brlO|n#j241NeEWaD||0hfIi({4oB|?YbXOTIpbZHf29$ zmkC8!Uf~Z2VmIjlbFkPQyFjeKCLDMV=%zpavJXeqbVZO&@LsrJ@f&p4&S%?#`P1y# zhm2W#Uabgfe+e8>Jy1t>9aG$lL|t)%y5P>&Vk5!`v9HqiPq*YQxyccer^*|~w0=jcSV4zXe>>bJ%7x>*Oduz$%3?#Kwx&{VNt6dJ z-@sWJQVGe^HPh&;Ec~Cvs}LJfzKPzz-QXg{1o|nG{l!zHv6n!iH_$L- zamZ~1CbfzWf>rJ1KdHj+zL)fd(7wCMUuM2x$hc6SrT+bs3E9SteY~4ZUG&85MT#6G+s4WX+1!ZZq6X6bmsR0c~}Rg-Cfneq?*{rUK&0E52=LxRP>NVSqsKghG~B(}EwqnirT zigP<#8?w8Mq}mBaaM=s{FsGc=(NaVoYZ2JLe(y%(5^pBlejk>F<3ft)QzS8QwW#jB zg{`p=v828AfcbFMlil$-r5d~GfJn(z(z@u7v z?>1mk_eRv1f4`y)OBhRu)Ih3gQr z{{wH5O#tfcgGm><%N}kxD~8+{u_Pa*5?d56?i80)G9mmN+1ba^Y;1M++t~;T26Ri> zC23=qfE44pfS5>n`A16^-ISL3-K}AiXX_mvRE8syb(r9MOFP1-5NFWoE&1%=%m$MS z?H{f_&))V}gdaAR~G(5X;K4T<{;iWO$^nRoE&7M^`30(1Mp&G55 zS^?BoV`PZu|7P9m$?wuG^KW@mxm5Y{8pSSTjNcp2t_AJFlIBt;+?E1I8qY@m-pwi~ zpd89Fff*oo|Bw}Of3CWx@%8(Mkq_j%!5++tBk| z@A^zTJdx$cogkl-e5NoA8kmK}@Gbbmn8%%}x|HG9AdsK-F#N6-*7-IhD;1Tf-qvKxL`e!H*FOs;R&}T7 z-P5Gu8;aY1#`D|ldUU%%cJ@^l!2sP2aS1yO1gF7m*-8K?=jg9H(0}5PYq)joD%Yn4 z^HCT5BnP!w1vJw)b?D4Ac(Q-a`1{%i1q~HrQ9mP-?Os7as^El7d(NQlsng`+7T&s8 z`@8bS-JdUEfDW3?HN0f0WE(Cr4E(`CMID znS@?!8{~NgF2jMC=d~ewt<2$@WBL`WK599d3gQv4mDywL%*0N>y--M$q2NWHsE89Y zvD4H<1nfbvme}qc09{9#n(l}>Rs$5QL*aCZ4MQ)&yr7TB$T?>czh@{gbEa4IJN~|9 z&}Iu9$jgCOcFzy)l~hnQ@upIm&Qh`lQ6@eiUAaQL{goQnAE9i-f{kPF14U1P?PMqI zucRSx6BCnvh4y`6{7)0pMZqpiy_c%yWDoJsgS<=F(*?qf#V8_?+*7Hwa=r@?ppB1=hM;^cHKX?`2(dz1nnI@XC@M2Rl@Q<>OfJoImuq78?C}3JxUs+ymky4;u zS(xvT_0339-j?8W6l#ZnW4d*eIA!`X9T#Ybcz+A4ECLhXOI=x`t!e%w^O z+$C{uyA={JX*4!IEE9?owlrm(S7B)FL4guN{jTT$LxXk{&yJlrchOk6+k)6m^8Dto zB0W#e8lgsHPGKK0_{yBSn^wQn!q|L4(;|-1h>6K|oXD2>RPQ0E!}1ro)%+;RBG7qwU+dSwlmol zzbXn)USJXwaDn^tIoU+p%J>W+g+EIs=ZP#J*pG(Sa`Ie+aiAgCQTrNKWbnu~?1KVv zGd(J)D{*^2k2(qWsKp+!*H*lND$E~O z>DiQ-rX0Sme90!Fr^vM)kxF~-*27x0=T9TV#!1*en$DQ&m;LVRvGe;d7~iCWbtZ+f zr66-(e&f}uc+XTA(fg{Gn8 z9dY)3-VKF!fd#0NNYm`^-^@MtUc6oul+&M-SEyXg+?-Gaa-i_1whj`t@$`Ji#^s3Z z+4l<#(bj#(gDbz=;xu?QRZ8Yn`kzlt?x*DnRooziJC*dO&Q>ue&+m9f@CtflKi}=+ zgY=?pnKMdfEr(zfc>g`f$F5jo``duYs=^I6ym-F%z!5XOho1krx) zG;U;&F&0sO7XJL&6owQqTRc0CL$W$hFUU75 z+l_%HVu|p74|d+Q`{W@@h>cVp?)fdy^@t7~sn5Q9mxE3Gf;~4g`s=6lQ&A#A-CAc< zU3yzV@iXjSs{ga%5;GoTf0bspW8gM6cr^Lg{S`{@o=>5GF?!EcvRY0-0+D*tn^Y7{ z)F(k1?>4FIA)`dl!yfO#04a~&R`g>#1pRB&d6;Fj%zmlX z{B2s`7NkGVWk(0*-S86hWcYb9#y50dDBahr^P3AXyCa{hIqp;Y6x!$AflMuJ`|5Yv zaH?`3v(t`HC@pKOO@o~)|1r(%b%vV{Ht`6F=F=1EYlZIq~gylT24g1s(E2&&qUv%6D#cR4A8(p;dF+>U+@FJ;cw z7GB`!zntCeDI-ziZHtN4}7pd2ZLZ$RziM^=c?QEJas@MC0!K^L@oii$*sS zp%e5om`Y;#%lz5q`*Xg1BMZ0IzqH~jE_s)iQxO8F`S+oNx!B#PiG>g%;c;XC`U0W6 z>}>vXc4227v3EcN@X0XMgsTt5v}SCem9S_%zW1dAnpe23W%_@zv0SIwFj4auGMB`; zL<#6`bA%nowRSibZ#vnqvOSSAG77>>9U)GXWK~|F2A(KvoXY@{#^wt-4cLF6eakj6 zuZQDQyf(4o

Oryy>2tGAxqSoA+n-%*uLX>pNlG;`h!XV%~9#pmAQbpn)KFer$)O zKv$a_e^w^yrj${Q)c{{Vz#5-^sivj#jR=4s}skk~K>XTu-*@P4(sHC8?OljwXgDNDRD zD;~zVnTI54qIJrCi&;E$va`_cZMy}F0-zxNi`l4(uRf21PNyCnef?KbRK&51GKLBa z+Sn)%;{Nn4tB-Y~_3@;MTD||6I+K11BqRf+a}@oW_Ye)=HFO-_clkGrwx(8^TpVXP zjTy6#4O=zkWuri{iySf+7ncWCR}k5LEY@a4XS;vmN9z~9qa{)Lfq8$$CmL|E$vv%{ zLI_2Envf@z?LzMiL5H`+}|ovjKJ5!>vxKg>}An8oN(L`sS0A1wXo_`IN2#r8g#Mgr`FR&2%xYV$f6V8{Wr z#+lP{zJO^4H7i<4H`;zCfM7?-m8mb>KM#!RDQI*6BU%*jMUNz4oai`+gBdG~1O+ z?J4opBzCN6{^6m2Y%Al=(V23uto_g$b7>#M6vmPYbb zhU{hak%Wu{s6lkcv@cNS+f$3lD8GD0)nv>Qtg`x2_$mexZJXorQdsP3-}U(b&%qB6 zLUW{IJECEUYEIrfwvQLX0*6snQPI7>Zz3GW{GDP`u3y`tXxfH6Ye+pCIA)K2d|%?p za4h!B>=ksQ%7i3e)O-lrQ}0VL#j=lI zR#*KgA0Q6rK@?fbg3P|J-q0V5JSjb|SxMJs|FgUa3mPdM!_IZ9%eFAbiCZUcRN zSdEBGNrkhV?af0~-%YNwq9bitOhg%!H{vgT709q>AR(Q6-6rfM;lxQHg?rvXHZl9U zFNlr_Z$(P-hnl}|1fzSOTQ#`p&T@Dy{;iQC+wQBNd*~Z5pO%`UZna|&Q+nV3E8e9T zbwkYK;?L?Ko7kI%vJPOG+V!}h!@To$w{*iQ=s>!j#7-~ehSxOj1#;xQ8ffcG^ZcjQ z1aX;7I@c(fgt=qgTNXQ>#Obo4*F{$Bo+>qw${Utm*NwV8o8omc* zHKG)cY%QNr?E)b0dZtfId05A6D#BXg{4nZ0X=}WTi03&+KFc|m9&dc}6{<*s3+SF% zac2C)gf+!8&phEGc1&6eBgREH@o=#qz9Ve$#3%c|UC3gft!MF9yVsx~9nj1gGkp5v z{poTYD){+<+*+8l%1i0O43?Ajz1*^y*uLBrhRA1O9{(kaW?9IFS?wt;0Q`%lqAW*p4uB&Gx z=r!-@Iky-v&|{*nzv<{ot%nbmlahHy!q&_;hVsai*Y>x%Vk5mst3L3fh~vhP2Tuq* z1Bs2o0pDj=k87j0aZG5(wys;>Z$!4;nA@AN+HR+O%2F(Y*%{kj<6Ht#^x?fb?wN4B zA%($|Gw1q?*eGm?5zTpr#x8PSG7^-B z=Xdqrc?oW=Za=#^D~|iKOJQc5^3lxgJuhy%oYseI<*x0X3T3*=NT%TQ$Ebt%?flk1Uo`96gA~8n|=mrwL%`9!5q9@YsDgU_4v0y$XFhb*dW!l$Twu~ zb<`VNjjtuI9My5n&CYgt0aw&_&UL^l49SVhL7H(e-b8ltv#vGd>|eg2#^c9(7GfOC zo>wAAWUPqUdwuO>taUpo@h!1+Q#Ew1c050o$uWv@A;3GzWDb8GmmW>n0bJxG&prU( zWtRvv-o4W8%Zn`dq-R|dmuwte1v^(p@-7fH(9-Iv^RwMG!3OudR|(sOsMiSx{Z*)N z<4Man<=&(-mZDalzkX`q@ykNGtwN8*^73e#LH99WEh}1Q7}w0z%NGtLh{W-??kzm8 z9zvb3Y7UwGPMazbCFMIkb&Prs7yT$>XF7|d!xg!kpaq0xC|7}RgNj+Ez6>O5ZY?C+ zdJ*(J-4VCFK!c>D*Y~>LL>M!5RdPe`63T^BDJRz<;4aGzE38GI>#}MZjpB_QnuBsQ z*>0ah4%zSAM|vAAzGeBB$du-6TF|UIW;7(mzQ)$dfEZ?-wiXg2VbABzB~|?fh)4ku z8EAYn?S>;}2wy7udcIe*Qy3(YG*KmrzXEZRMB_xWe_QEy+0jq`rtdCu?azYLUc|^W z9P_>%Ahy5sMgV0^)XPxL*p_w z6i=rmj*z(dX8z|ALxt}1j#Nqzo-DOIXk|3_O=QL=jat0hDr9A?kp|cug!qe0gP%(g zk^OLeV2CN29^IcmE9UIV`!k!(5>g|C%_?xH{U z3`%2&Cp#Ls^LOX@faHCbsX*_ER?!PML+cCE4SnTHQ5_m&WwJV0NGp!H<1f9zD;(^K zANAL3Y@-^JaC~KL-R(~%@_qW-XWz8jl2e-cEE%VQ5xvDnv_$t++93IEZ}guC9X`oI zS$U(vxQP6lS#Kz=l~N2x|2swTmCL!uMxfb!X1_zCurr{%MS(03YbV|e{tyS6ys$(r zlihP}3Zf8&^FVC=^{UW(!NvQ7pZ~A^F{snlZy85s;4N0rQ@~@*MDKBCKD&maD z4KvZFjjccOb*qgw{RYVtKxo?vF%K zA}Pwoul38OlI5cK!|Y%Lr+@Ck{nEhM9?MFZ%u%OaIsTo=kpp5ob=My)-}RQ=`tnrM ztfDwS60|`OUh!`INn|UQjh=GbmL=JLjP~!V)yMa@_kkfdk^?BCUWze*<5q%F5Zj(& zcw@jR;a5{y61zluWAo}$!Fms_M;)K}Bl)$38Amt2%s}$wQhm;_Dx60CMKfm))xv?0 z$o`w*dxd|B_bctvsH+lj(^l;xliLbm zn4DY0qa`aY%)Q=m0IfX6n(=f!ER?(H#%W)o687`slPBkeQn~<4aGlk8d!aKU3goQW zO$K8%xz>o{)|fMZYlZr z>%4{1PS|4RCzbHRvQ00OgSNrEvbF@C0}qPf0rwH6rfi7N0vIGnZ)hX#6k}J9OzB~H z3VEXJK;@EK6zGoIh&(#PZJ!f|`1owJ_{3fRb#yn?9NwK?Hi+GP=SU9JeJ=F9gAh`o z5_bwHlKohj3&PepZW*&HZ5;U@B^^CFY3c)8`kx-eTdlM>&?eN;pLF{F&&!|E&_{H z{uRKrVfdE8*ipyJ)M|}~BGzRcjWlqVCLNWr{jxi1IpTgdrI-1?*9y5O3;Mp|gb=F6 zk5QAUEl|jms|mZ}?tM%vx(m#G+=+!XrGmAsKC$xM@-MZgGYR6P1J-qu-Sb^O3s;^) z_dYzLRfs8-wHetf7fN|_vHPX>XE#z!TbFhw1y}K}COQ)8Agofn*k_n+weJcLJxBt2 zUQFLrue^n50nZ-*Xf0gB-s`heo)M6fV}6}= zsr343b%?n_d?aV}g&c_7Ab+A2V9sd15q-k7X1~w1?h+^-n#{JFO;qV1xU$k|8 zsc>UPSP^s|N#&eVMsb$^r@brxhq`~)*|L_gH!?(aqQ{c0GL~Y5DB04UQl1t|lrkEr z>|`f0p{VRjRJKZqkw!hC9#NK>B2h{t&h_b>AI=|eemFmT`z7gR#%JE|_x--_>%Ok* zzR4AGj!Yj_9;A139S)zWYi&s1W^vVWav5Rl*pi!ZX@MtJw@{=jLs74nZ~swKZzAa!%lw3X6bk3YMaJdMt6!0HLk95 zvEY|Cjg&Y8+t~VVAH9!?3yDM%vsFT;6?lc#mI{Ydow617T@FK>=dKzkVjk1Z^8>1=s~pswProQnB53RqB|SyozVZ z%AXn*o9lzA=Hc^Ycwo4{6K<$d^8A0a)HU8LX)`6?=4Ui{LL(UD)d7@Sw=%i#N=SSB z$_SGG;j7f-s&F)-Omqk`Hsym24y{TkQh~9$Y#JKe!?ELaLrcwcG�UE=(LV)A23Z zc1xN*UjMCApJ(gVtswQBn?O(6=HV{0y-ebv%Ktd+(FSgA(=yXJ2z{%9 zNj?A`c610Dr(*P%+uS_fS*{V|fG)Fw_Z`^7zhS%lw5Yt;>y94s>fUYe&myQvHV3(0 zIWR#XjQ<@gXYxFLMFdHFxulQ3E6J64AGAg&!=Zov%S@@Y5*94O!^60Rw!gz_ov3V} zo`dBWVHE|vX!(Po-M~M3U*dd@MHVaG6AT207#tzRh-}_JuXmA}yDo6F`)QjXcTPat zQMj6&mob&}u@o>7RA&ztpXH{Uhpg9Fl2fd_U2qEGMCD&ssS=j9eNN1_wy!YuR=qnn zu<$#VA`=C%;Wfk`5SJ$mdw|})fvT+c2j*%RJVZ*m4#a}(#~R=b{~W`dz42?ju07TS zUqJo1<9L*r@33ML3WJ|&6>@737~pMw2nQF-x{oIP$BgzHp#0sU7)Z^*X;EUpARzrs zBiytP41TzBdMN{STJr*?+Ur4^2cFD~os8UPf-j@GhG<;ierr$HO-MoW1;eIH3J%9G z=BxcdvEi3LxyXEaFhGLnD~^VT*XVcZQ=yoaJtIvIE6OjXP{hFQB@Y)g{y){<}5AFHeG4d^uCFR57!N2Dq= zg(_3|>>q@T3hOC5mv^9oMf2^sz383v{TAGD4Q=7efy^54{RZ#n(p1_$3wN~ zlVwbr6Z|?2@)l63Sa>#j^97xN364VS znUr4jT=}$%PxhcESaIsYfio~YSmRhTeeUWDk9(;NqVrUa5&_s|&#g;$<|n(XrN{Q(ZRVgHk?IT*v>8ii&!o9H6iPNi7!4-3Dj|ii7mNcf=X;Q#~8RvlIVy6bWGZ6(RQP=kLzeypc<>rye_&lCNp zzoafHS)NIW9cLU-zP+VxWTZ`1dboPDu04~`_tgh+feKTs;&%!HYbBMOF?~=8r81jgOl!F~S zr*IFB)fpTghIA*mFG0Zaug^ZBD8i$Y$vXCE`|WiI_%7v>qW;P}WG{%+$rmn;H2xY- z&bj@6@Na{>ex?f`V^L{vB+2MWa#l>O!Q|FLiVe088QaGsMRQxbpMub3 zhMv_v=#`MJH-~}ZbI6m)c@Mr?i6U;oc(e58?1DXChm$wkNtEgbZe#9Y^iNMnZeo6a z2IADev>kcEfjDd(RK9h*uMT6)#q~fe6ZEJ(x?{FlZ6`CO2xr>ES#`D@^Pp}Stk%$I z^Q$_F@UJ%mOZPjnw@Qn@2xl?;gsYEjwpvw!bqMm3CFtq}hx+U9AU`|9edKAjSmGh^ zn8xhIvqHqU5oKapv+OedXKTO3jvd5BExRnA%JNrlo(}CR2i>5s0E3@u$nv$+(XW&! zE`<{1h{P&>A%iwK`P0B#DC{?lq4y#p*W5S6lEJ;lH8YwiT?cn^Yej3BZPahthgRGW z9~_A@>Q7WXgF%zxdH5XyYMl0g|2zPgY}JBMVYqvX8vE0@xSYG($-Jvvi!){f*2nOAZk_-mp=RiL1{~K4CJ%(` zW}*QV^@L`AeN>A*`dD!p2|MY?_tV zbJS4UiOD47+ZWdmS+Bq$=;iJtzffmGev)p0u59(t9Gd-az!?-YxcGVByZnUp=>GPC z^%+y?T(Nh<%*H5*bSi7^BQy}N#B|6)Q=;Z`t6}RV`5oEA@0n*|-A1xQ(i}ue=C^Z{ z{-A%4)P;>`BwL}NAz}TrQ7fp*cY~oSDE65cvp1))RSc<>Zdm*3@_F)KE@MO*K?N-3 zW+$8DNO5Ie{SJO`M)9K{fP$X7)f)JmCw81lU6QhDyY%TIr>o&}r_kH_4Z^5?&|AWY zmJ(d-sY+rApTcXT^#TwYW~GZNRwh2Vs6baYtL>IDiQ_IgcMt9(7bqWE4~cQWne*aq zIQC^GJZCLxs7lj|;iarLf$9#g+MQIFJOKmxrrASw5Q1#p$n1(Kax1$s4k~k30TS6r zONMIZMtBEx)k$l*2pm;>*alOH?(@Tenh%f%1|dWI_RL&i;R`0X99`O#!i)5lO^rbs zV=EiFYjrV|CFhewQioHci$13~J zPaBqD7tX5Y2LH8w7anFbMB{F6faG}70=<=z;y|NSmkk`@pxxeAwnZyUOekm4>S%$t34rEU$m`enX zyoLyGyR%qunk`!Llwd=H-v!6d0IKQiJc-=adqlXv5#g!ex? zcbStJ-P^qIzwvbnjhEcFGQC4Y!Ou>{WjrYu2$K|*?%6Z@BDfs{6c;0+XFjRTx&_D) zS|LK5kt-nfsvSC=we$3`iq?+aNHaGJ`HfcZT`ozHUoM6$yQjI1&vdX;^AhAF7TsoIqPd$e;&aSeh#d_53?I~O*tIJq8Ad!=N>wVu1NB-*Djm@EX1KT%nF1ioM z;&=a6MRu=s(hgY5b+4F0N$Y*d2K2tso^2hQgM=%C3WmgdLg)86GPeY-3fd9v+_gNB zcvj+3p)I$~a#!1JXzD@sT)o7qizW*>C8NT^1I7*EDS!-`wEf-S+5OE-Mr|egEyA{x zH)8T#jT}7X6N$-HyV}!MnPD#3K(LsIqgU-$`j5@77o^YpR*j~P9?aX533kD`r2fDc z4!Dz2d+o(INCG}xJH}?-hX9Aji@ZGKQfp_1U;*%qWt^>j}0g{ z>;6!)@-<1Glx*Y!n*>Bo!y7AoN6@P~AFgPfjR>!MC?NPfj5t1cLrjaSJciNw{yJ43ibfzcdG#e9aH)h4n_mbe&z*y`eoC?+53DlUgb!<98TTkH65hOi(k>Kkgf~=)S_@^!UAE?P_D;q z*7n4wf4X#8xI&$Rrc{(GXef=l1`@OX;*&LUCi3jv#HUv}xMTnk2?7Pf>TrG)*?RzkswW{K^LS)@VMFt=b z!{_sh{Y876TUmY#<|E|CHH}|#l9IRsQVW%SPG49KVXM`FLU{S{ni(}&XJlt*5C2UW7+Y59obmbPmU7;Rk% zJ1gYw5B$_NaAa7iuzVWze}MS!s!d>d=mfJ6aT-C!n{JWnbY{i44M2z P_(R*^XkBW#JN&-@C8n8< literal 0 HcmV?d00001 diff --git a/docs/404.html b/docs/404.html index 34285ba..1a78c50 100644 --- a/docs/404.html +++ b/docs/404.html @@ -4,18 +4,18 @@ - Page not found | Clinical Data Analysis for Cancer Informatics + Page not found | Appendix I - + - + @@ -57,28 +57,6 @@ div.hanging-indent{margin-left: 1.5em; text-indent: -1.5em;} - @@ -158,11 +136,31 @@

  • 5 Clinical Data Management
  • 6 Appendix I: Ontologies, Vocabularies, Taxonomies, and Standards
      @@ -190,7 +188,7 @@
      diff --git a/docs/clinical-data-management.html b/docs/clinical-data-management.html index 881ebf4..1aedea1 100644 --- a/docs/clinical-data-management.html +++ b/docs/clinical-data-management.html @@ -4,18 +4,18 @@ - Chapter 5 Clinical Data Management | Clinical Data Analysis for Cancer Informatics + Chapter 5 Clinical Data Management | Appendix I - + - + @@ -57,28 +57,6 @@ div.hanging-indent{margin-left: 1.5em; text-indent: -1.5em;} - @@ -158,11 +136,31 @@
    • 5 Clinical Data Management
    • 6 Appendix I: Ontologies, Vocabularies, Taxonomies, and Standards
        @@ -190,7 +188,7 @@
        @@ -214,21 +212,148 @@

        Chapter 5 Clinical Data Management

        5.1 Learning Objectives

        +

        Learning Objectives: Learning Objectives: 1. Understand the clinical data handling process and how to choose clinical data handling tools, 2. Explain privacy considerations for clinical data, 3. List major regulations and regulatory bodies for clinical data, 4. Describe different types of documents used when working with clinical data

        5.2 Clinical data handling tools

        +

        Clinical research relies heavily on the collection, processing, analysis, and management of data. Efficient and effective data handling is crucial to ensure the integrity, reliability, and validity of clinical trials and studies. Selecting the right tools for handling clinical data is a critical step in the research process. This section defines data handling in the context of clinical research and contrasts it with data quality, discusses the importance of privacy, provides a comprehensive guide on how to find suitable clinical data handling tools with a focus on open-source options, and emphasizes the importance of testing data handling methods using dummy data in the preliminary phases of a clinical trial.

        +
        +

        5.2.1 Clinical data handling

        +

        Clinical data handling involves the processes and tools used to collect, manage, store, process, and share clinical data in a way that maintains its integrity, security, and usability. It encompasses a range of activities, including:

        +
          +
        • Data Collection: Gathering data from various sources such as case report forms (CRFs), electronic health records (EHRs), patient surveys, and laboratory results.
          • +
        • Data Storage and Management: Safely storing data in databases or data warehouses and managing access to ensure that only authorized personnel can interact with the data.
          • +
        • Data Cleaning and Processing: Ensuring data consistency, accuracy, and completeness by detecting and rectifying errors, missing values, and inconsistencies.
          • +
        • Data Analysis: Using statistical and analytical tools to generate insights from the data that can inform clinical decisions and study outcomes.
          • +
        • Data Sharing and Reporting: Providing access to data for collaborators, regulators, or stakeholders in a secure and controlled manner and generating reports that summarize findings.
          • +
        +

        The goal of effective data handling is to ensure that data is accurate, reliable, and available when needed, while also protecting patient privacy and complying with regulatory standards.

        +
        +

        5.2.1.1 Data quality vs data handling

        +

        Data quality and data handling are closely related but distinct concepts in clinical research.

        +
          +
        • Data Quality: Refers to the accuracy, completeness, consistency, and reliability of data. High-quality data is essential for producing valid and reliable research outcomes. It involves processes such as data validation, error checking, and quality control measures.
          • +
        • Data Handling: Encompasses the broader scope of managing the data lifecycle, from collection and storage to processing, analysis, and sharing. While data quality is a component of data handling, the latter also involves aspects like data security, access management, and regulatory compliance.
          • +
        +

        Both data quality and data handling are critical for ensuring the integrity of clinical research. Data handling tools must support high data quality standards through features like automated data validation checks, error reporting, and data cleaning functions.

        +
        +
        +
        +

        5.2.2 Selecting clinical data handling tools

        +

        Choosing the right tools for clinical data handling depends on several factors, including the study’s size, complexity, data types, regulatory requirements, and budget. Key considerations include:

        +
          +
        • Compatibility and Integration: The tool should be compatible with existing data systems and workflows, allowing for seamless integration of data from multiple sources.
          • +
        • Scalability: The tool should be able to handle the volume of data expected in the study and scale as the study progresses.
          • +
        • User-Friendliness: A user-friendly interface can reduce the learning curve for researchers and data managers, improving efficiency and reducing errors.
          • +
        • Regulatory Compliance: The tool should comply with relevant data privacy and security regulations to protect patient data.
          • +
        • Support and Documentation: Adequate support, including user manuals, tutorials, and customer service, is essential for troubleshooting and maximizing the tool’s potential.
          • +
        +

        Research teams should conduct thorough evaluations, including reading reviews, seeking recommendations, and conducting pilot tests, to identify the best tool for their specific needs.

        +
        +

        5.2.2.1 Open source options

        +

        Open-source tools provide a cost-effective and flexible alternative to proprietary software for handling clinical data. These tools are often developed and maintained by vibrant communities and can be customized to fit specific research needs.

        +

        Advantages of open-source tools include:

        +
          +
        • They are typically free to use
          • +
        • They provide transparency by allowing users to review the codebase for security and compliance
          • +
        • They can be modified to meet unique requirements
          • +
        • They often have strong community support, which can be valuable for troubleshooting and enhancing functionalities.
          • +
        +

        Examples of popular open-source data handling tools include:

        +
          +
        • REDCap: A secure, web-based application for building and managing online surveys and databases, widely used in clinical research for data collection and management.
          • +
        • OpenClinica: An open-source clinical trial software that supports data capture, management, and analysis, compliant with regulatory standards like FDA requirements for electronic data (FDA 21 CFR Part 11).
          • +
        • KNIME: A data analytics, reporting, and integration platform that can be used for data cleaning, processing, and analysis, with extensive capabilities for machine learning and data visualization.
          • +
        • R and Python: These programming languages offer powerful libraries and packages (such as the tidyverse for R, and pandas and NumPy for Python) that support a wide range of data handling and analysis tasks.
          • +
        +

        Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform that offers some basic statistical and data visualization tools, and offers APIs to allow the real-time analytics in other platforms (R, C#, cURL, etc.). However, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued guidance on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR.

        +

        While open-source tools are beneficial, it is essential to ensure they are secure, well-maintained, and compliant with relevant privacy and regulatory standards.

        +

        When moving between multiple data handling tools, it is important to consider interoperability. As an example, REDCap and R play very nicely together via API tokens, but this may not be true of all electronic data capture systems or databases. The sharing, integration and redistribution of data between systems - whether internally or externally - needs to be carefully planned, tested and documented.

        -
        -

        5.3 Importance of keeping clinical data safe

        +
        +
        +

        5.2.3 Testing data handling tools

        +

        Before implementing data handling methods in an actual clinical trial, it is crucial to run tests using dummy data. This step ensures that the data handling process is robust, efficient, and free from errors across different conditions, without risking sensitive patient information.

        +
          +
        • Creating Realistic Dummy Data: Dummy data should mimic the real data in terms of format, structure, and complexity. It should include various scenarios (e.g., missing data, outliers, data entry errors) to test the system’s error-handling capabilities.
          • +
        • Simulating the Full Data Handling Workflow: The process should involve every stage of data handling, from data collection and entry to storage, cleaning, analysis, and reporting. This comprehensive simulation helps identify potential issues early, such as data loss, security vulnerabilities, or errors in data processing algorithms.
          • +
        • Refining and Optimizing Data Handling Methods: Based on the findings from dummy data tests, researchers can refine their data handling protocols, adjust tools and settings, and optimize workflows to ensure smooth operations when real data is introduced.
          • +
        +

        Testing with dummy data provides an additional layer of quality assurance and helps build confidence in the data handling process before the clinical trial begins.

        +
        +
        +

        5.2.4 Conclusion

        +

        Effective clinical data handling is fundamental to the success of clinical trials and studies. Selecting the right tools requires careful consideration of privacy, regulatory compliance, open-source versus proprietary options, and the specific needs of the study. Understanding the difference between data quality and data handling is crucial, as both are essential for ensuring the validity and reliability of research outcomes. Running preliminary tests using dummy data is a critical step in validating data handling methods, ensuring that the chosen tools and processes are robust, secure, and efficient. By following these guidelines, researchers can enhance data management practices, protect patient privacy, and achieve meaningful and reliable clinical research outcomes.

        +
        +
        +
        +

        5.3 Privacy considerations for clinical data

        +

        Privacy is a fundamental consideration in clinical data management, given the sensitive nature of the information involved. Clinical data often contains personally identifiable information (PII) or personal health information (PHI), which must be protected to comply with privacy regulations such as the Health Insurance Portability and Accountability Act (HIPAA) in the United States, the General Data Protection Regulation (GDPR) in the European Union, and other regional laws. More information about PII and PHI can be found in this course about ethical data handling.

        +
          +
        • Data Anonymization and De-identification: One of the primary methods to protect privacy is to anonymize or de-identify data, removing or encrypting any information that could directly or indirectly identify an individual.
          • +
        • Privacy-Preserving Record Linkages: In scenarios where data from multiple sources need to be linked without compromising individual privacy, privacy-preserving record linkage techniques are essential. These methods enable the integration of datasets by matching records in a way that minimizes the risk of re-identification. Techniques such as secure multi-party computation, homomorphic encryption, and differential privacy can be employed to ensure that the linkage process itself does not expose sensitive information.
          • +
        • Access Control and Encryption: Tools should support robust access controls, ensuring that only authorized users have access to the data. Encryption should be used for both data at rest and data in transit to prevent unauthorized access.
          • +
        • Compliance and Auditing: Tools should facilitate compliance with regulatory standards and provide auditing capabilities to track data access and usage.
          • +
        +

        When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research.

        5.4 Government regulators

        +

        Clinical data is governed by several different types of regulations. In this section, we will review some of the major regulatory frameworks and organizations.

        +
        +

        5.4.1 Health Insurance Portability and Accountability Act (HIPAA)

        +

        The Health Insurance Portability and Accountability Act (HIPAA), regulated by the U.S. Department of Health and Human Services (HHS), establishes national standards to protect individuals’ medical records and other personal health information. It applies to health plans, healthcare clearinghouses, and healthcare providers that conduct certain healthcare transactions electronically. The HIPAA Privacy Rule requires appropriate safeguards to protect the privacy of protected health information (PHI) and sets limits on the uses and disclosures of such information without patient authorization.

        +
        +
        +

        5.4.2 Food and Drug Administration (FDA)

        +

        The U.S. Food and Drug Administration (FDA) regulates the safety, efficacy, and security of human and veterinary drugs, biological products, medical devices, food, cosmetics, and products that emit radiation. The FDA’s regulations ensure that clinical trials are conducted ethically and that data collected is reliable and accurate. This includes oversight of clinical trial protocols, informed consent, and reporting of adverse events.

        +
        +
        +

        5.4.3 General Data Protection Regulation (GDPR)

        +

        The General Data Protection Regulation (GDPR), governed by the European Union (EU), is a comprehensive data protection law that governs the collection, processing, storage, and transfer of personal data within the EU. It aims to enhance individuals’ control over their personal data and simplify the regulatory environment for international business. GDPR applies to any organization that processes the personal data of EU residents, regardless of where the organization is based. Key provisions include the right to be forgotten, data portability, and mandatory breach notifications.

        +
        +
        +

        5.4.4 Honest brokers

        +

        Honest brokers act as neutral intermediaries between the data source and researchers, typically regulated by Institutional Review Boards (IRBs) or equivalent bodies. They are responsible for de-identifying data to ensure that researchers cannot trace the data back to individual patients. Honest brokers must complete specific training, such as Collaborative Institutional Training Initiative (CITI) Research Ethics and HIPAA training, before accessing data. They play a crucial role in maintaining the confidentiality and integrity of clinical data.

        +

        5.5 Documentation

        +

        In data management, several types of documentation are frequently encountered, each serving a specific purpose throughout the lifecycle of a research project. These documents are crucial for ensuring data integrity, regulatory compliance, and effective project management.

        +
        +

        5.5.1 Data Management Plan

        +

        A Data Management Plan (DMP) outlines how data will be handled during and after a research project. Typically created at the beginning of a project, it is stored in project documentation repositories or institutional databases. Often required by funding agencies, a DMP may include plans for testing with dummy data to ensure data integrity. The importance of a DMP lies in its role in planning and managing data throughout the project, ensuring that data is handled consistently and responsibly.

        +
        +
        +

        5.5.2 Statistical Analysis Plan

        +

        The Statistical Analysis Plan (SAP) details the statistical methods and analyses to be performed on the collected data. Developed before data analysis begins, it is stored with project documentation or in electronic lab notebooks. Essential for clinical trials and other research requiring rigorous statistical analysis, the SAP may involve testing statistical methods on dummy data. The SAP is crucial for maintaining the integrity and reproducibility of statistical analyses.

        +
        +
        +

        5.5.3 Standard Operating Procedures

        +

        Standard Operating Procedures (SOPs) provide detailed instructions on how to perform specific tasks or processes. Used throughout the project lifecycle, they are stored in organizational repositories or document management systems. SOPs are often required for regulatory compliance and may include procedures for testing with dummy data. The importance of SOPs lies in their ability to standardize processes, ensuring consistency and compliance with regulatory requirements.

        +
        +
        +

        5.5.4 Data Use Agreements

        +

        Data Use Agreements (DUAs) define the terms and conditions for data sharing and use. Encountered before data is shared with external parties, they are stored in legal or administrative offices. DUAs are required when sharing data with external collaborators but are typically not directly related to dummy data. DUAs are important for protecting data privacy and ensuring that data is used appropriately.

        +
        +
        +

        5.5.5 Data Sharing Agreements

        +

        Data Sharing Agreements (DSAs) specify the terms for sharing data between organizations. Encountered prior to data sharing, they are stored in legal or administrative offices. DSAs are necessary for formalizing data sharing arrangements and generally do not relate to dummy data. The importance of DSAs lies in their role in facilitating collaboration while protecting data integrity and compliance with legal requirements.

        +
        +
        +

        5.5.6 Documentation across the span of a project

        +

        These documents are encountered at various stages of a project, from planning (DMP, SAP) to execution (SOP) and data sharing (DUA, DSA). They are typically stored in project documentation repositories, institutional databases, or document management systems, with access restricted to authorized personnel. Not all documents are required for every project; their necessity depends on the project’s scope, regulatory requirements, and institutional policies. Some documents, like the DMP, SAP, and SOP, may include provisions for testing with dummy data to ensure data integrity and validate processes.

        +

        To ensure compliance with these documents, organizations should implement robust data governance frameworks that include regular audits, training programs, and clear policies and procedures. Standardized templates for these documents are often available from funding agencies, regulatory bodies, or institutional guidelines, helping to ensure consistency and compliance with best practices. It’s best to ask before starting any document to ensure the correct format is used.

        +

        For further reading, consider the following resources:

        +
          +
        • Guide to Clinical Data Management Procedures (GCDMP): GCDMP
          • +
        • Books by Suzanne Prokscha: Writing and Managing SOPs for GCP and Practical Guide to Clinical Data Management
          • +
        +

        Additionally, remember the insightful quote by Damian Conway: “Documentation is a love letter that you write to your future self.” This is an invaluable tidbit to keep in mind throughout the lifecycle of a study.

        +
        -
        -

        5.6 Conclusion

        +
        +

        5.6 Conclusion

        diff --git a/docs/reference-keys.txt b/docs/reference-keys.txt index 72187e0..7d9769d 100644 --- a/docs/reference-keys.txt +++ b/docs/reference-keys.txt @@ -57,3 +57,24 @@ rxnorm loinc emerging-standards-such-as-fhir omop +clinical-data-handling +data-quality-vs-data-handling +privacy-considerations +selecting-clinical-data-handling-tools +open-source-options +testing-data-handling-tools +conclusion-2 +hipaa +fda +gdpr +honest-brokers +health-insurance-portability-and-accountability-act-hipaa +food-and-drug-administration-fda +general-data-protection-regulation-gdpr +data-management-plan +statistical-analysis-plan +standard-operating-procedures +data-use-agreements +data-sharing-agreements +documentation-across-the-span-of-a-project +privacy-considerations-for-clinical-data diff --git a/docs/search_index.json b/docs/search_index.json index 20ed5e0..95f3996 100644 --- a/docs/search_index.json +++ b/docs/search_index.json @@ -1 +1 @@ -[["index.html", "Clinical Data Analysis for Cancer Informatics About this Course", " Clinical Data Analysis for Cancer Informatics May, 2026 About this Course This course is part of a series of courses created for the Informatics Technology for Cancer Research (ITCR) Training Network (ITN). The ITN is a collaborative effort of researchers around the United States that supports cancer informatics and data science training through resources, technology, and events. This initiative is funded by the following grant: National Cancer Institute (NCI) UE5 CA254170. Our courses feature tools developed by ITCR Investigators and make it easier for principal investigators, scientists, and analysts to integrate cancer informatics into their workflows. Please see our website at itcrtraining.org for more information. Except where otherwise indicated, the contents of this course are available for use under the Creative Commons Attribution 4.0 license. You are free to adapt and share the work, but you must give appropriate credit, provide a link to the license, and indicate if changes were made. Sample attribution: Cancer Data Management by Fred Hutchinson Data Science Lab (CC-BY 4.0). You can download the illustrations by clicking on this link for our slides. "],["introduction.html", "Chapter 1 Introduction 1.1 Motivation 1.2 Topics Covered 1.3 Curriculum", " Chapter 1 Introduction 1.1 Motivation This course is intended for researchers (including postdocs and students) with limited to intermediate experience with informatics research. The conceptual material will also be useful for those in management roles who are collecting data and using informatics pipelines. This course is intended to provide an overview of the different kinds of data commonly used in clinical research, the questions that can typically be asked with such data, as well as guidance for how to manage this data. 1.2 Topics Covered 1.3 Curriculum The course will cover key underlying principles and concepts in ethical data handling. "],["clinical-data-types.html", "Chapter 2 Clinical Data Types 2.1 Learning Objectives 2.2 Clinical data is unique 2.3 Major clinical data types 2.4 How to acquire clinical data 2.5 Description of data 2.6 Summary", " Chapter 2 Clinical Data Types Clinical data is health-related information collected from patients throughout their healthcare journey. It may come in many forms and its sensitive nature requires careful management by researchers. 2.1 Learning Objectives 2.2 Clinical data is unique Clinical data refers to information collected from patients during healthcare delivery, clinical trials, and medical research. Clinical data comes from a wide variety of sources and as such, requires careful consideration when designing, collecting, and analyzing this data. Unlike domains such as Finance or other areas in the sciences which predominantly use structured data with predictable and consistent formats, clinical data is often heterogeneous, integrating many forms of both structured and unstructured data: quantitative measurements, categorical data, genotyping, images, subjective narratives from patient notes, and objective observations or conclusions. The unstructured nature of free text from notes reflecting subjective patient experiences or qualitative insights from healthcare professionals especially adds a further layer of complexity. Furthermore, the contrast between patient and doctor notes reflects the dual perspectives of symptoms and formal diagnoses. In essence, clinical data’s unique blend of structured and unstructured components, along with its multidisciplinary nature, necessitates specialized methodologies for comprehensive analysis and interpretation in the realm of healthcare. Further, because clinical data contains sensitive, personal information about patients, there are additional security and ethics concerns in the handling and management of clinical data. 2.3 Major clinical data types Clinical data can come in many different forms, including patient demographics medical history or records such as diagnoses, lab results, vital signs, medication records, or procedure history genetic reports health monitor data images scanned documents, and notes written by physicians, nurses, and other clinicians survey/ case report form (CRF responses) and more … Some sources of clinical data are more prevalent and readily obtainable than others. For instance, notes, demographics, images, and histories or observations/records (lab results, vitals, medications, procedures) are often stored directly in electronic medical record systems making them more easily accessible and so are the focus of most Electronic Health Record (EHR) data research efforts. 2.3.1 Structured data Observational records such as test results and demographic data are often collectively referred to as “structured data”, as they are stored in electronic health record databases and often provided to researchers in tabular form. Structured data types frequently used in EHR research consist of comprehensive longitudinal records of a patient’s interactions with a healthcare system and may include demographics, diagnoses, lab values, procedures, vitals, and medication records. The structured nature of this data allows for it to be stored in tables which may be indexed by a patient or visit ID and often include timestamps and other supporting descriptors. For example, medication orders might specify the drug name, class, dose, unit, quantity, route, frequency, and other instructions. Structured data tables often describe entries in terms of codes from standardized vocabularies. Diagnoses might be described with codes from the International Classification of Diseases (ICD) vocabulary, lab tests with Logical Observation Identifiers Names and Codes (LOINC), medications with National Drug Code (NDC), and procedures with Current Procedural Terminology (CPT) codes. These terms, or “billing codes”, provide a common foundation that can be invaluable for identifying patients with a specific disease or who have received specified medications, particularly when integrating data from multiple sources. As we’ve described it, structured clinical data is expected to have similarities, although specific details may vary based on the type of EHR being used and any customizations to the EHR for the specific environment or institution in which data was collected (e.g., any specialized pre-processing by institutional research offices prior to providing data to researchers). So it is important for researchers to consider differences present in their data if they’ve obtained it from multiple contexts or institutions. 2.3.2 Unstructured data / clinical notes Clinical notes are, perhaps unsurprisingly, generally shared as seemingly straightforward text files. However, the simple format should not be taken as a suggestion that the data are easy to interpret. Some EHR systems contain literally dozens of types of notes, covering specialties such as pathology or surgery; specific moments in care such as admission or discharge; particular procedures such as colonoscopies; patient-provider interactions such as telehealth or phone encounters, and many others. In addition to differing in content, these sources may have different layouts and formats, ranging from free-form reports to structured SOAP (subjective, objective, assessment, and plan) formats or even templated procedure reports. Understanding the types of notes available in a given context and where relevant data might be found is a key step in effectively using clinical notes. When used in EHR research, both structured data and clinical notes are generally de-identified to protect patient privacy. Patient ID numbers might be replaced with new identifiers, with linkages maintained by institutional “honest brokers” (Dhir et al. 2008) charged with providing clinical data for research purposes. In some cases, dates may be changed as well. Clinical notes are generally “de-identified” through specialized software designed to remove names, dates, locations, and other sensitive details. Researchers working with institutions to access clinical data should be sure to understand local data de-identification practices. 2.4 How to acquire clinical data 2.4.1 Secondary Sources 2.5 Description of data 2.5.1 File types 2.5.2 Metadata 2.6 Summary References "],["specific-clinical-data-types.html", "Chapter 3 Specific Clinical Data Types 3.1 Learning Objectives 3.2 Physiological 3.3 Monitoring data 3.4 Radiology 3.5 Pathology 3.6 Synthetic Data 3.7 Summary", " Chapter 3 Specific Clinical Data Types 3.1 Learning Objectives 3.2 Physiological 3.3 Monitoring data This section was written by: Jennifer Kelleher, Ph.D.1; Abigail S. Robbertz, Ph.D.1; and Meghan E. McGrady, Ph.D.1,2 NOTE: Jennifer Kelleher, Ph.D.1 and Abigail S. Robbertz, Ph.D.1 contributed equally. 1 Center for Adherence and Self-Management, Division of Behavioral Medicine and Clinical Psychology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA 2 Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH, USA The work discussed in this section was also supported by the National Cancer Institute at the National Institutes of Health (R21CA263704, K07CA200668) to MEM. JK and ASR are supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development at the National Institutes of Health (T32HD068223). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Electronic monitoring devices are digital tools that can be used to track health behaviors over time such as: Sleep Physical activity Medication adherence Calorie intake Electronic monitoring devices can also be used to assess physical health indicators including: * Blood glucose levels * Blood pressure * Heart rate and heart rate variability * Oxygen saturation Electronic monitoring devices enable researchers to track day-to-day health behaviors in the patient’s “real-world” setting. This allows researchers to explore patterns or changes in a patient’s health behavior and provides a richer understanding of daily behavior over time. 3.3.1 Benefits of Monitoring Data Electronic monitoring devices often include data transmission abilities that enable healthcare providers or researchers to access these data in near real-time potentially informing intervention and/or medical decision-making. Electronic monitoring devices also have the potential to produce more accurate estimates of health behaviors than alternative strategies (e.g., self-report) as they are not subject to recall bias and can detect efforts to inflate adherence due to social desirability. 3.3.2 Considerations This section is not exhaustive. Research teams are strongly encouraged to consult with experts with experience and training in collecting and analyzing data from specific devices. To ensure the outcome variables are aligned with the research question of interest and ethical and age/developmental considerations (Psihogios et al. (2024); Modi et al. (2012)) have been appropriately accounted for, readers are encouraged to consult with researchers in their field who have integrated these measurement strategies into their work. 3.3.2.1 Medication Adherence There are three major components of medical adherence (the tracking of taking medication): Initiation: Starting a prescribed regimen Implementation: The amount of which a patient’s medication-taking behavior corresponds with the treatment regimen or protocol Discontinuation: Stopping a perscribed regimen For more information see: A new taxonomy for describing and defining adherence to medications (Vrijens et al., 2012) Pediatric self-management: A framework for research, practice, and policy. (Modi et al., 2012) 3.4 Radiology 3.5 Pathology 3.6 Synthetic Data 3.7 Summary References "],["clinical-data-uses.html", "Chapter 4 Clinical Data Uses 4.1 Learning Objectives 4.2 General uses of clinical data 4.3 Types of questions that can be asked with clinical data 4.4 Research Design Considerations 4.5 Conclusion", " Chapter 4 Clinical Data Uses 4.1 Learning Objectives 4.2 General uses of clinical data 4.2.1 Cancer Prevention and Care The near universal adoption of electronic health record (EHR) systems in the US has created unprecedented opportunities to improve cancer prevention and care. As described in previous chapters, EHR systems store comprehensive longitudinal records of a patient’s interactions with a healthcare system, including data about: demographics conditions family history clinical notes laboratory test results medications procedures imaging genetic test reports These data can be accessed not only by health professionals, but also by patients through patient portals. EHR data can also be used to enable data-driven interventions such as provider- and patient-facing clinical decision support (CDS) and population health management (PHM). CDS has been defined as tools that “provide clinicians, staff, patients, or other individuals with knowledge and person-specific information, intelligently filtered or presented at appropriate times, to enhance health and health care” (Osheroff et al. 2007). Examples of widely adopted CDS tools with demonstrated effectiveness for cancer prevention, diagnosis, and care include: provider and patient reminders for cancer screening decision aids and shared decision-making tools (e.g., for prostate cancer treatment, lung cancer screening) AI-supported diagnostic imaging chemotherapy decision support at-home symptom care While CDS tools generally provide decision support focused on a specific patient at a time, PHM are strategies that target specific patient populations (Swarthout and Bishop 2017). PHM efforts generally consist of: population algorithms that are applied over EHR and other data sources to identify individuals who are eligible for a specific healthcare intervention (e.g., colorectal cancer screening, tobacco cessation, HPV vaccination) patient engagement strategies (e.g., patient portals, text messaging, chatbots, patient navigators) that provide education and access to those interventions analytic tools that assess the effectiveness of the PHM program. Several PHM programs have demonstrated to be effective in increasing the uptake of cancer prevention. For example, the colorectal cancer screening program at Kaiser Permanente uses digital (i.e., text messaging, patient portal), mailed, and patient navigation approaches to increase colorectal cancer screening by mailing Fecal Immunohistochemical Test (FIT) kits to patients’ homes (Gupta et al. 2020). Also the Cancer Moonshot BRIDGE trial used the GARDE platform (Bradshaw et al. 2022) (ITCR-funded) to identify candidates for genetic testing of hereditary cancer syndromes based on EHR data; and for patient outreach, pre- and post-test education via automated chatbots (Kaphingst et al. 2024). Clinical Decision Support (CDS) can help patients and clinicians make decisions about an individual’s care, while Population Health Management (PHM) can help identify individuals for interventions and engagement. The image shows a single person getting a colorectal screening reminder for CDS and a population being identified for possibly needing colorectal screening for PHM. While some CDS and PHM approaches have been successfully adopted widely, emerging technologies such as the use of generative AI approaches to analyze diagnostic imaging, large language models (LLMs) to extract information from narrative texts (e.g., clinical notes), LLM-based chatbots to communicate with patients, and digital health tools such as home-based sensors are creating unprecedented opportunities for next generation CDS and PHM. These approaches have the potential to enable significant breakthroughs through the implementation of patient-tailored cancer prevention and care at a population scale. Nevertheless, substantial research is needed to ensure effective and fair implementation of these CDS and PHM interventions. 4.3 Types of questions that can be asked with clinical data 4.3.1 Risk Prediction Risk prediction in clinical research involves using data to assess the likelihood of certain outcomes or events occurring in patients. This could include predicting the risk of developing a particular disease, experiencing a specific complication, or responding to a treatment. Data used for risk prediction can come from various sources, including: Clinical Data: This includes patient demographics, medical history, laboratory results, and imaging studies. Genetic Data: Genetic information, such as DNA sequencing results, can provide valuable insights into an individual’s susceptibility to certain diseases. Environmental and Lifestyle Data: Factors such as diet, exercise habits, smoking status, and environmental exposures can influence disease risk and may be included in risk prediction models. Biomarkers: Biological markers indicative of disease or physiological processes can be used as predictors in risk models (Bodaghi, Fattahi, and Ramazani 2023). Once relevant data is collected, statistical and machine learning techniques can be applied to develop predictive models. These models aim to identify patterns and relationships within the data that are associated with the outcome of interest. Common techniques include logistic regression, decision trees, random forests, support vector machines, and neural networks. After the model is trained on a dataset, it can be validated using independent datasets to assess its performance and generalizability. Once validated, the model can be used to predict risk in new patients based on their individual characteristics and data. Clinical prediction rules are a subset of risk prediction models, specific to clinical research. Examples of risk prediction models … Risk prediction models are important because … Overall, risk prediction in clinical research allows healthcare professionals to identify individuals at higher risk of certain outcomes, enabling targeted interventions, personalized treatments, and more efficient resource allocation. 4.3.2 Cohort identification for research Clinical data plays a crucial role in cohort identification for research purposes. Researchers typically use electronic health records (EHRs), medical databases, or registries to identify cohorts based on specific criteria such as age, gender, medical conditions, treatments, medications, and outcomes. Advanced data mining and natural language processing techniques can also be employed to extract relevant information from unstructured data sources like clinical notes. Once cohorts are identified, researchers can analyze the data to study disease progression, treatment effectiveness, and outcomes. Cohort identification is important regardless of research study type, but to provide specific examples: Research Design: Identifying cohorts allows researchers to design studies with appropriate inclusion or exclusion criteria. By selecting specific groups of individuals with similar characteristics or exposures, researchers can investigate hypotheses effectively. Clinical Insights: Cohort studies enable researchers to observe the natural history of diseases, track outcomes over time, and assess the effectiveness of interventions or treatments. Understanding how different factors influence disease progression or treatment response can inform clinical decision-making and improve patient care. Epidemiological Studies: Cohort identification is crucial for epidemiological research to understand the incidence, prevalence, and risk factors associated with diseases. By following cohorts over time, researchers can identify trends, patterns, and associations that contribute to our understanding of disease causation and prevention. Precision Medicine: Identifying cohorts based on genetic profiles, biomarkers, or other specific characteristics allows researchers to tailor treatments and interventions to individual patients. This approach, known as precision medicine, aims to optimize therapeutic outcomes while minimizing adverse effects. Healthcare Policy and Planning: Cohort studies provide valuable data for informing healthcare policies, resource allocation, and public health strategies. By identifying high-risk populations or groups with specific healthcare needs, policymakers can develop targeted interventions to improve health outcomes and reduce disparities. 4.3.3 Case report forms (CRFs) In clinical research, case report forms (CRFs) are essential tools for collecting standardized data from study participants. Note that case report forms are any paper or form that will be filled in at the case or participant level. By that definition, even a consent form is a case report form! And each clinical study may utilize multiple CRFs (e.g., one for consent, another for medical history, another for reporting any adverse effects). CRFs are useful for several tasks already discussed within this chapter – specifically, tracking adverse events or outcomes (like those discussed in the Risk Prediction section) or tracking demographics and medical history for identifying cohorts (as discussed in the Cohort Identification section). CRFs are also useful for topics this chapter will discuss in later sections (e.g., Retrospective analyses). Designing CRFs that are accessible and sensitive to the different needs of participants requires careful consideration. Questions within CRFs should be formulated to gather comprehensive and accurate clinical data while ensuring participants feel safe, respected, and comfortable. This section explores the types of questions that can be asked using CRFs with a focus on accessibility, sensitivity, specificity, and comfort for people While within a study, CRFs help to ensure standardization in data collection, there may be a lack of standardization when comparing data between studies if each study did not use the same CRFs or comparable/subsets of questions within forms. This section will provide guidelines for types of questions that may be found within CRFs and writing these questions; however, there aren’t necessarily templates that have been used widely within the field. Further, your specific study needs may also require different or additional types of questions – the guidelines within this section are not exhaustive/all-encompassing of what may be encountered within the field. Here, we explore the different categories of questions that may be included within case report forms: Demographic and Socioeconomic Questions Capturing demographic and socioeconomic data is fundamental in clinical research to understand the background of study participants. However, these questions must be asked in a manner that respects privacy, avoids assumptions, and helps identify people with different life experiences. It’s also good practice in order to respect privacy to include an option allowing the participants to leave the question unanswered or respond with “prefer not to answer”. Sex and Gender: When asking about gender and sex include an option of “prefer to self-describe” to capture more information. Ethnicity and Race: Ethnicity and race questions should be specific and use respectful language, allowing participants to self-identify rather than selecting from a predefined list. Socioeconomic Status: Questions about employment, income, or education should be framed to capture social determinants of health without making participants feel judged. For example, asking, “What is your current employment status?” with choices that include full-time, part-time, unemployed, student, and unable to work can help gather relevant data without stigma. It is essential to use neutral, non-judgmental language and to explain why these questions are being asked, ensuring participants understand the relevance of their responses. Health and Medial History Questions Health and medical history questions provide critical information about baseline conditions and potential risk factors of participants. These questions should be framed clearly and respectfully to avoid any discomfort. Medical Conditions and History: Questions about past and present health conditions should use clear, accessible language. For example, “Have you ever been diagnosed with any of the following conditions? (Please check all that apply)” followed by a comprehensive list containing all necessary options. Medication Use: Questions about current and past medications should include over-the-counter and alternative therapies, and space should be provided for free-text responses to capture additional details. Disability and Functional Status: For many populations, it is important to use person-first language, such as “Do you have any physical, sensory, or cognitive impairments that you would like us to be aware of?” and provide space for participants to describe their specific needs. However, different populations have different preferences. The “Disability Language Style Guide” from the National Center on Disability and Journalism provides some basic guidelines, thorough discussion, and community specific advice on this topic (“Disability Language Style Guide” 2021). Avoiding medical jargon and providing definitions or examples can help ensure that participants understand the questions, and confidentiality should be emphasized to encourage honest responses. Experience and Quality of Life Questions Understanding how health conditions and treatments affect participants’ daily lives and well-being is essential, particularly for those from communities who experience health disparities and may also experience unique challenges. Daily Living and Social Functioning: Questions like “How often do your health conditions affect your ability to perform daily tasks (e.g., cooking, cleaning, working)?” can help assess the impact on daily life, with options ranging from “never” to “always.” Emotional and Psychological Well-Being: Including questions such as “In the past week, how often have you felt anxious or depressed?” using a scale from “not at all” to “very often” can provide insights into mental health needs. Support Systems and Social Networks: Asking about social support (e.g., “Do you have someone you can rely on for emotional support?”) can help identify participants’ needs for social and emotional resources. Using sensitive language and providing mental health support resources where needed is crucial when discussing emotional well-being to avoid triggering emotional distress. Sexual and Reproductive Health Questions Questions about sexual and reproductive health must be asked with sensitivity, as they can be deeply personal, particularly for groups who may face stigma. Sexual Orientation and Gender Identity (SOGI): Instead of just predefined categories, open-ended questions like “How would you describe your sexual orientation?” and “What is your gender identity?” allow participants to self-identify. Offering the option to skip these questions respects participants’ privacy. Options can be provided with a space or text box for the person to fill in their own descriptor, but having the categorical data will allow for easier analysis for those that select one rather than every response being from variable open text responses. Reproductive Health: Questions about menstrual health, contraception, or pregnancy should be framed neutrally. For example, “Are you currently using any form of contraception? If yes, please specify.” Sexual Activity and History: Questions should be direct but framed sensitively, such as “Are there any sexual health concerns you would like to discuss? Your answers will help us understand how to better support your care needs.” These questions should always be optional, with confidentiality emphasized to encourage honest, comfortable participation. Treatment Preferences and Decision-Making Questions Understanding participants’ preferences for treatment and decision-making is vital for providing patient-centered care, especially for certain groups of people. Decision-Making Preferences: Questions like “How involved would you like to be in decisions about your healthcare?” offer a range of choices from “I prefer to make decisions myself” to “I prefer my healthcare provider to make decisions,” allowing participants to express their autonomy. Cultural and Religious Considerations: Asking, “Are there any cultural, religious, or personal beliefs that we should consider when discussing treatment options with you?” ensures that care is respectful and culturally appropriate. Treatment Burden: Questions such as “What level of inconvenience or side effects would be acceptable to you when considering a treatment?” help to gauge participants’ preferences and comfort levels. These questions should be framed to respect participants’ autonomy and encourage honest responses without fear of judgment. Accessibility and Accommodation Needs Questions To ensure that all participants can fully engage with the study, it is essential to ask about accessibility and accommodation needs. Language and Communication Needs: “What is your preferred language for communication? Do you need an interpreter or translated materials?” These questions help ensure that participants can understand the materials. Physical Accessibility: Asking, “Do you require any specific accommodations to participate in this study (e.g., wheelchair access, hearing aids, visual aids)?” ensures physical accessibility. Format Preferences: “Would you prefer to complete this form online, on paper, or verbally with assistance?” helps accommodate different needs and preferences. Providing multiple options and allowing participants to request changes at any time is crucial to accommodate evolving needs. Providing multiple options and allowing participants to request changes at any time is crucial to accommodate evolving needs. Cultural Sensitivity and Identity Questions CRFs should respect different cultural backgrounds, values, and identities without perpetuating biases or assumptions. Cultural Identity and Practices: An open-ended question such as “Are there any cultural practices or beliefs that are important for us to be aware of in your care?” allows participants to share relevant information. Dietary Restrictions and Preferences: Asking, “Do you have any dietary restrictions or preferences that are culturally or religiously motivated?” ensures that these are respected. Community and Belonging: “Is there anything about your community or background that you would like us to know to provide better care?” encourages participants to share relevant aspects of their identity. These questions should be open-ended, allowing participants to skip questions they find irrelevant or uncomfortable. Ensuring Comfort, Trust, and Privacy Questions Fostering a sense of safety and trust is especially important for individuals who may have experienced discrimination in healthcare settings. Comfort and Confidentiality: Asking, “Do you feel comfortable with the way your information is being collected and stored? Are there any specific concerns you would like to address?” helps build trust. Feedback and Preferences: “Is there anything about this form or the study process that you find confusing, uncomfortable, or concerning?” invites participants to share their feedback. Consent and Voluntary Participation: Questions like, “Would you like to be contacted about the results of this study or for future research opportunities? Participation is entirely voluntary.” reinforce autonomy and respect. Reminding participants of the confidentiality and voluntary nature of their involvement can help foster a trusting environment. The guidelines above sometimes suggest use of open-ended questions where participants would provide free-text responses rather than selecting pre-defined categories. This will require researchers to process those free-text responses and may decrease the overall standardization. Designing specific and sensitive CRFs for clinical studies requires a thoughtful approach that respects the different backgrounds, privacy, and comfort of all participants, especially those from communities that experience health disparities. By using accessible language, offering multiple options, respecting autonomy, and providing a safe space for participants to express themselves, researchers can gather meaningful and accurate data while ensuring participants feel valued and respected. These considerations are vital to fostering clinical research that captures enough information about a wide variety of individuals. Case report forms (CRFs) are often tailored to specific studies and may vary widely in structure and content, lacking standardization across different projects. Despite this, certain themes are commonly expected in CRFs, including sections on participant demographics, medical history, treatment outcomes, and adverse events, ensuring essential data collection across a variety of study designs. By thoughtfully addressing both unique study requirements and universally relevant data points, researchers can optimize CRFs for consistency across clinical studies. 4.3.4 Clinical studies and trials Clinical data generated from clinical trials and observational studies form the backbone of evidence-based medicine. The ability to analyze and interpret this data enables researchers and clinicians to answer a variety of important questions that directly impact patient care, treatment decisions, and healthcare policies. Clinical trials are tightly regulated studies, controlling patient recruitment and monitoring administration and impact of treatments or interventions (“What Are Clinical Trials and Studies?” 2023). The goals of clinical trials include assessing the safety and efficacy of new treatments or interventions, comparing different/existing treatment options, determining the optimal dosage and administration of interventions, and identifying potential side effects or adverse events. Definition 4.1 The NIH’s definition of a clinical trial is “a research study in which one or more human subjects are prospectively assigned to one or more interventions (which may include placebo or other control) to evaluate the effects of those interventions on health-related biomedical or behavioral outcomes.” The NIH provides 4 questions and additional clarifications to consider when identifying if a study is a clinical trial or just a clinical study (“NIH’s Definition of a Clinical Trial,” n.d.). Definition 4.2 Prospectively assigned: Refers to a pre-defined process (e.g., randomization) specified in an approved protocol that stipulates the assignment of research subjects (individually or in clusters) to one or more arms (e.g., intervention, placebo, or other control) of a clinical trial (“NIH Clinical Trials,” n.d.). Here, we explore the different categories of questions that can be addressed using clinical trials data, ranging from the evaluation of treatment efficacy and safety to the exploration of predictive factors for disease prognosis and the personalization of medical care. Efficacy and Effectiveness Questions One of the primary objectives of clinical trials is to determine the efficacy of new treatments or interventions. Clinical data is used to answer fundamental questions such as “Is the new treatment effective?” and “How does its effectiveness compare to existing treatments?” Researchers design studies to measure the magnitude of treatment effects, comparing outcomes between treatment and control groups (e.g., a new drug versus a placebo or standard care). Beyond controlled settings, data from real-world evidence (RWE) studies can further address questions about the effectiveness of treatments when applied in routine clinical practice. For example, “Is Drug A more effective than Drug B in managing hypertension in a real-world patient population?” Such questions help to assess how treatments perform outside the controlled environment of clinical trials where patient adherence comorbidities, and polypharmacy may influence outcomes. While clinical trials are an important source of real-world data (RWD), they are typically conducted under tightly controlled conditions, which may not fully reflect the complexities of real-world studies. This makes RWE particularly valuable for understanding how treatments work when patients have varying levels of adherence or multiple chronic conditions that are often excluded from clinical trails. Definition 4.3 Comorbidities: refer to the presence of one or more additional medical conditions or diseases that coexist with a primary condition in a patient. These conditions can occur independently or be related to the primary disease. Safety and Tolerability Questions Safety is a critical aspect of any clinical study, and clinical data is essential in identifying and characterizing the safety profile of new treatments. Questions like “What are the common adverse effects of the new treatment?” or “What is the incidence of serious adverse events (SAEs) in patients receiving Drug A?” are addressed by carefully monitoring and recording adverse events throughout the study. Comparative safety questions also arise, such as “Does Drug A have fewer adverse effects compared to Drug B?” These questions are crucial in weighing the benefits of a treatment against its risk and in identifying specific populations that may be at higher risk of adverse reactions. Definition 4.4 Serious Adverse Events (SAEs): Any medical occurrence during a clinical trial or medical treatment that results in significant negative outcomes, such as death, life-threatening situations, hospitalization (or its prolongation), persistent or significant disability, or congenital anomalies. SAEs also include any event that requires intervention to prevent one of these outcomes. Definition 4.5 Incidence: Refers to the rate or number of new cases of SAEs that occur within a specific population during a defined time period, typically during a clinical trial or study. Is it usually expressed as a proportion, such as the number of SAEs per 100 or 1,000 participants, and helps measure how frequently these serious events happen among those exposed to a particular treatment or intervention. Comparative Effectiveness and Cost-Effectiveness Questions Beyond safety and efficacy, clinical data can be used to evaluate the comparative effectiveness of different treatments and their cost-effectiveness. These questions often guide healthcare policy and clinical guidelines: “Is Drug A more cost-effective than standard care for managing chronic heart failure?” and “What is the cost per quality-adjusted life year (QALY) gained with the new intervention?” Such analyses provide valuable information for stakeholders, including healthcare providers, payers, and policymakers, to make informed decisions on resource allocation. Definition 4.6 Cost per quality-adjusted life year (QALY) is a metric used in health economics to assess the value of a medical intervention by measuring the cost of gaining one year of life adjusted for its quality. A QALY incorporates both the quantity and quality of life lived, where one QALY is equivalent to one year of life in perfect health. If a treatment improves both the length and quality of life, it earns more QALYs. The cost per QALY is calculated by dividing the cost of the treatment by the number of QALYs gained, helping to determine the cost-effectiveness of healthcare interventions. Lower cost per QALY indicates better value for money in terms of health benefits achieved (2015). Mechanistic and Biomarker Questions Clinical studies often explore not just whether a treatment works, but also how it works. Mechanistic questions delve into the underlying biological pathways affected by a treatment. For instance, “Does Drug A work by inhibiting a specific enzyme involved in the disease process?” Biomarker data can play a key role in such analyses, helping to identify molecular markers that predict response to treatment. Questions like “Are there biomarkers that can predict which patients are more likely to benefit from Drug A?” help in developing targeted therapies and personalized treatment approaches. Predictive and Prognostic Questions Clinical trials and studies frequently aim to identify factors that predict disease outcomes or treatment responses. For example, “What baseline characteristics predict a better response to Drug A?” or “What are the predictors of mortality in patients with severe heart failure?” Such prognostic and predictive questions are crucial for identifying high-risk patients, guiding treatment decisions, and developing clinical guidelines. Identifying patient subgroups that benefit more or less from a treatment also facilitates personalized medicine approaches, ensuring that the right treatment is delivered to the right patient at the right time. With the advent of personalized medicine, clinical data is increasingly being used to answer questions tailored to individual patient profiles. For instance, “What patient characteristics modify the effect of the treatment?” or “Can we predict which patients are most likely to benefit from Drug A using machine learning models?” These questions are at the forefront of precision medicine, where the goal is to customize healthcare to each patient based on their unique genetic, biomarker, and clinical characteristics. Quality of Life and Patient-Reported Outcome Questions In modern clinical research, the impact of treatments on patients’ quality of life and their subjective experiences has gained prominence. Questions such as “How does Drug A affect patients’ quality of life compared to standard care?” or “What are the patient-reported outcomes associated with the new intervention?” are increasingly addressed in clinical trials. Data from quality of life assessments which are standardized and validated questionnaires (for example, the Health Assessment Questionnaire or HAQ), patient-reported outcome measures (PROMs, tools or questionnaires used to assess a patient’s health status or health-related quality of life directly from the patient’s perspective), and other patient-centric endpoints (Smoking History, Surgical History, etc.) provide valuable insights into how treatments affect patients beyond traditional clinical endpoints. Longitudinal and Follow-Up Questions Long-term follow-up studies are essential for understanding the durability of treatment effects and any delayed adverse effects. Questions like “What are the long-term outcomes associated with Drug A?” and “Is there a sustained benefit of Drug A in reducing symptoms after five years?” are vital for assessing the overall value of treatments. Such longitudinal analyses help in determining the optimal duration of therapy, the need for maintenance treatments, and the overall benefit-risk profile of an intervention. Safety in Special Populations Questions Clinical data can also be used to evaluate the safety and efficacy of treatments in special populations, such as children, pregnant women, or elderly patients. Questions such as “Is Drug A safe for use in pregnant women?” and “What is the risk of adverse events when Drug A is administered to patients with multiple comorbidities?” are essential for developing age- and condition-specific clinical guidelines. These questions help ensure that treatments are tailored to the needs of different patient populations, minimizing harm and maximizing benefits. Adherence and Compliance Questions Patient adherence to prescribed treatment regimens can significantly impact the outcomes of clinical studies. Questions like “What factors influence patient adherence to the new treatment?” and “How does adherence affect treatment efficacy?” are addressed through analyzing adherence data through monitors if a regulated study, or with interim/baseline analysis if non-regulated. Understanding these factors can lead to interventions that improve adherence and, consequently, the effectiveness of the treatment. Real-World Evidence and Generalizability Questions Real-world evidence (RWE) studies help bridge the gap between clinical trial results and clinical practice. Questions such as “How generalizable are the results of this clinical trial to the broader population?” or “What is the impact of real-world use patterns on treatment outcomes?” are essential for understanding how well clinical trial findings apply to everyday clinical settings. These questions help determine if the benefits observed in clinical trials can be replicated in diverse patient populations under routine care conditions. Prevention and Risk Reduction Questions Clinical data is also crucial in answering questions related to prevention and risk reduction strategies. For example, “Can the new treatment reduce the risk of developing diabetes in high-risk individuals?” or “What factors are associated with a reduced risk of cardiovascular disease in a large cohort study?” These questions are fundamental in preventive medicine, guiding public health interventions and informing clinical practice. Clinical Data provides a wealth of information that can be leveraged to answer a broad array of questions in clinical trials and studies. From understanding treatment efficacy and safety to exploring long-term outcomes, quality of life impacts, and the potential for personalized medicine, clinical data is foundational to advancing medical knowledge and improving patient care. Case report forms (CRFs) are instrumental in identifying patient cohorts or subgroups, documenting baseline characteristics, and capturing information on comorbidities, while also monitoring adherence data, which are essential for ensuring accurate and meaningful analysis. The type of questions that can be asked and answered are continually evolving as new data sources, analytical methods, and research paradigms emerge, further enriching the field of clinical research. 4.3.5 Retrospective analysis Retrospective analysis involves the examination of pre-existing clinical data to answer specific research questions, explore hypotheses, and identify patterns or trends. This method leverages historical data collected from medical records, administrative databases, registries, electronic health records (EHRs), or other sources of clinical information. The types of questions that can be asked through retrospective analysis span a wide range of clinical and epidemiological domains. The questions typically focus on understanding patient characteristics, disease epidemiology, treatment outcomes, risk factors, healthcare utilization, and more. Here, we explore the different categories of questions that can be addressed using retrospective data analysis: Descriptive Questions Descriptive questions aim to summarize and describe the characteristics of a patient population, disease, or healthcare process. They provide a foundational understanding of a dataset and are often the starting point for more complex analyses. Patient Demographics and Characteristics: What are the demographic profiles (age, gender, ethnicity, socioeconomic status) of patients diagnosed with a specific condition? What are the common comorbidities and risk factors in this population? Disease Prevalence and Incidence: What is the prevalence or incidence of a specific disease or condition in a particular population or geographic area over a defined period? Clinical Presentation: What are the most common presenting symptoms, clinical signs, or laboratory findings associated with a specific disease or condition? These questions help in understanding the baseline characteristics and epidemiology of diseases, which is crucial for planning further studies or developing public health interventions. Treatment and Intervention Questions Retrospective data can be used to examine the real-world effectiveness, safety, and patterns of treatment and interventions. These questions are critical for understanding how treatments are applied in clinical practice and their outcomes: Treatment Effectiveness: What is the effectiveness of a specific treatment or intervention in reducing symptoms, improving quality of life, or achieving clinical outcomes in a real-world setting? Adherence and Persistence: What are the rates of adherence and persistence with a prescribed treatment regimen over time? What factors are associated with higher or lower adherence rates? Comparative Effectiveness: How does the effectiveness of one treatment compare to another in a similar population? What are the relative benefits and risks associated with different treatment options? By examining historical data, researchers can gain insights into how different treatments perform outside the controlled environment of clinical trials, thereby informing clinical decision-making and guidelines. Outcome and Prognostic Questions Understanding patient outcomes and prognostic factors is central to retrospective analyses. These questions focus on the end results of healthcare practices and patient management, including survival, complications, and quality of life: Survival and Mortality: What are the survival rates and mortality rates associated with a particular disease, condition, or treatment? What factors are associated with increased or decreased survival rates? Complication Rates: What are the rates and types of complications associated with specific diseases, surgeries, or treatments? Are there identifiable risk factors for these complications? Prognostic Factors: What are the key prognostic factors (e.g., age, stage of disease, comorbidities) that influence the outcomes of patients with a particular condition? These questions help identify factors that influence patient outcomes, guiding clinicians in predicting disease progression and tailoring individual treatment plans. Risk Factor and Predictive Modeling Questions Retrospective analyses are often used to identify potential risk factors for diseases and develop predictive models. These questions aim to uncover associations between variables and outcomes to help predict future events: Risk Factors for Disease: What are the risk factors associated with developing a particular disease or condition? Are there demographic, genetic, behavioral, or environmental factors that significantly increase the risk? Prediction Models: Can we develop a predictive model to estimate the likelihood of disease progression, relapse, or adverse outcomes based on historical patient data? Multivariable Analysis: How do multiple factors interact to influence the risk of an outcome? Are there synergistic or antagonistic effects between different risk factors? These questions are critical for developing tools that help clinicians assess risk and make more informed decisions in preventive care and early intervention. Healthcare Utilization and Cost Questions Understanding how healthcare resources are utilized and the associated costs is important for optimizing healthcare delivery. Retrospective data can provide valuable insights into patterns of care and resource allocation for questions such as: Healthcare Resource Utilization: What are the patterns of healthcare utilization (e.g., hospitalizations, emergency room visits, outpatient visits) for patients with a specific condition? How does utilization vary by patient demographics, disease severity, or geographic location? Cost of Care: What are the direct and indirect costs associated with managing a specific disease or condition? How do different treatment options impact overall healthcare costs? Cost-Effectiveness: How does the cost-effectiveness of one treatment or intervention compare to another? What are the economic impacts of different clinical management strategies? These questions can inform healthcare policy, budget planning, and the development of strategies to improve cost-effectiveness and access to care. Quality of Care and Patient Safety Questions Retrospective analysis can be used to evaluate the quality and safety of care provided to patients, identifying areas for improvement and potential patient safety concerns, helping to answer questions such as: Quality Indicators: Quality indicators are standardized measures of health care quality and may focus on more specialized groups of care like preventative, inpatient or emergency care, pediatric or maternal health, etc. What are the levels of observed mortality or morbidity for a certain population in a specific setting? How utilized are certain procedures? (“AHRQ Quality Indicator Measures,” n.d.) Adverse Events and Safety: What is the incidence of adverse events, medication errors, or healthcare-associated infections in a particular patient population? What factors are associated with higher or lower rates of adverse events? Patient Satisfaction and Experience: What are the reported levels of patient satisfaction with healthcare services? How does satisfaction vary by demographic or clinical characteristics? These questions help in assessing healthcare quality, identifying gaps in care, and developing interventions to enhance patient safety and satisfaction. Temporal and Trend Analysis Questions Retrospective studies often involve analyzing temporal patterns to understand how diseases, treatments, or outcomes change over time. These questions help in recognizing trends and patterns that can inform future research and policy decisions: Temporal Trends in Disease Incidence and Prevalence: How have the incidence and prevalence of a specific disease changed over the past decade? Are there seasonal or cyclical patterns in disease occurrence? Trends in Treatment Practices: How have treatment practices and prescribing patterns evolved over time? Are there emerging trends in the use of certain medications or interventions? Trends in Outcomes and Survival: How have patient outcomes, such as survival rates or complication rates, changed over time with advances in treatment and healthcare delivery? These questions are essential for monitoring public health trends, evaluating the impact of interventions, and planning future healthcare strategies. Subgroup and Stratified Analysis Questions Subgroup analyses can help uncover differences in disease characteristics, treatment responses, and outcomes among different patient groups. This type of analysis is particularly valuable for personalized medicine and addressing questions related to health disparities: Differences by Demographic Group: How do treatment outcomes, risk factors, or disease characteristics differ between various demographic groups (e.g., age, gender, race/ethnicity)? Subgroup-Specific Effectiveness: Does a particular treatment work better in one subgroup of patients compared to another? What factors might explain these differences? Health Disparities: Are there disparities in healthcare access, treatment outcomes, or disease burden across different socioeconomic or minority groups? Questions focusing on subgroups are crucial for understanding variations in care and outcomes, identifying at-risk populations, and developing targeted interventions. Retrospective analysis of clinical data allows researchers to ask a wide range of questions that provide valuable insights into patient characteristics, disease epidemiology, treatment effectiveness, outcomes, risk factors, healthcare utilization, and more. By carefully formulating questions and analyzing historical data, researchers can uncover patterns, identify trends, and generate evidence that informs clinical practice, healthcare policy, and future research directions. This chapter outlines the various categories of questions that can be addressed through retrospective analysis, highlighting the potential of this approach to advance knowledge and improve patient care in diverse clinical settings. 4.4 Research Design Considerations Researchers need to intentionally use methods earlier in the research process than data analysis to manage data biases associated with clinical data, especially EHR. One of the most important challenges in using EHR data in cancer research is that, as in many other fields, healthcare data are plagued with several types of biases that result from disparities in the delivery of healthcare. Overall, cancer research has historically relied on data from high resource academic medical centers, which disproportionately provide care to patients who are White, have high socioeconomic status, and live in urban areas. As a result, medical knowledge produced from these data have disproportionately benefited those patients. Different sources of bias are prevalent in EHR data. For example, Information representativeness bias occurs when certain groups are disproportionately less present in the EHR because they have no contact with the healthcare system. Information presence bias, on the other hand, occurs when certain groups may be represented in the EHR, but have disproportionately less comprehensive healthcare data due to issues such as overall lower access to and use of healthcare service, lack of a primary care provider, lack of access to specialty care, and lack of access to digital resources (e.g., patient portals, home sensors, telehealth) that can be used to provide healthcare data. Treatment biases happen when certain groups receive disproportionate access to more advanced treatments, which is often determined by social drivers such as insurance, distance, health literacy, and socioeconomic status. Algorithm bias further amplifies these previous sources of bias by leveraging biased EHR data to make predictions about diagnosis, treatment and prognosis that are used by clinicians to make potentially biased healthcare decisions, which are then documented in the EHR. Recent advances in sophisticated and costly technology such as genetic testing, artificial intelligence, and digital health, which are disproportionately available in high resource healthcare systems further compound the problem. Therefore, cancer researchers increasingly need to use intentional methods to prevent, identify, and correct for biases in EHR data. For example, the National Institutes of Health Pragmatic Trials Collaboratory has made several recommendations to address EHR data biases in research (Andrew D. Boyd et al. 2023a; Andrew D. Boyd et al. 2023b): Include data from low resource healthcare settings such as community health centers that provide care for patients who have low socioeconomic status and live in rural areas. Engage with communities during study design and study conduct to ensure proper data collection, analysis, and representation. Use data collection methods for self-reported data that rely on more accessible technology such as text messaging, using accessible and culturally adapted communication. Include subgroup analysis by different demographic groups according to variables such as socioeconomic status, race, ethnicity, sex, geographical location, and social determinants of health. With such approaches, cancer researchers while aiming to avoid exacerbating health disparities, in addition help to reduce disparities. 4.5 Conclusion References "],["clinical-data-management.html", "Chapter 5 Clinical Data Management 5.1 Learning Objectives 5.2 Clinical data handling tools 5.3 Importance of keeping clinical data safe 5.4 Government regulators 5.5 Documentation 5.6 Conclusion", " Chapter 5 Clinical Data Management 5.1 Learning Objectives 5.2 Clinical data handling tools 5.3 Importance of keeping clinical data safe 5.4 Government regulators 5.5 Documentation 5.6 Conclusion "],["appendix-i-ontologies-vocabularies-taxonomies-and-standards.html", "Chapter 6 Appendix I: Ontologies, Vocabularies, Taxonomies, and Standards 6.1 SNOMED 6.2 ICD10 6.3 RxNorm 6.4 LOINC 6.5 Emerging standards such as FHIR 6.6 OMOP", " Chapter 6 Appendix I: Ontologies, Vocabularies, Taxonomies, and Standards 6.1 SNOMED 6.2 ICD10 6.3 RxNorm 6.4 LOINC 6.5 Emerging standards such as FHIR 6.6 OMOP "],["about-the-authors.html", "About the Authors", " About the Authors These credits are based on our course contributors table guidelines.     Credits Names Pedagogy Lead Content Instructor(s) FirstName LastName Lecturer(s) (include chapter name/link in parentheses if only for specific chapters) - make new line if more than one chapter involved Delivered the course in some way - video or audio Content Author(s) (include chapter name/link in parentheses if only for specific chapters) - make new line if more than one chapter involved If any other authors besides lead instructor Content Contributor(s) (include section name/link in parentheses) - make new line if more than one section involved Jennifer Kelleher, Ph.D.1*; Abigail S. Robbertz, Ph.D. Content Editor(s)/Reviewer(s) Julia K. Herriott, B.S. Content Director(s) Helped guide the content direction Content Consultants (include chapter name/link in parentheses or word “General”) - make new line if more than one chapter involved Gave high level advice on content Acknowledgments Gave small assistance to content but not to the level of consulting Production Content Publisher(s) Helped with publishing platform Content Publishing Reviewer(s) Reviewed overall content and aesthetics on publishing platform Technical Course Publishing Engineer(s) Helped with the code for the technical aspects related to the specific course generation Template Publishing Engineers Candace Savonen, Carrie Wright Publishing Maintenance Engineer Candace Savonen Technical Publishing Stylists Carrie Wright, Candace Savonen Package Developers (Leanbuild) John Muschelli, Candace Savonen, Carrie Wright Art and Design Illustrator(s) Created graphics for the course Figure Artist(s) Created figures/plots for course Videographer(s) Filmed videos Videography Editor(s) Edited film Audiographer(s) Recorded audio Audiography Editor(s) Edited audio recordings Funding Funder(s) Institution/individual who funded course including grant number Funding Staff Staff members who help with funding   ## ─ Session info ─────────────────────────────────────────────────────────────── ## setting value ## version R version 4.3.2 (2023-10-31) ## os Ubuntu 22.04.4 LTS ## system x86_64, linux-gnu ## ui X11 ## language (EN) ## collate en_US.UTF-8 ## ctype en_US.UTF-8 ## tz Etc/UTC ## date 2026-05-10 ## pandoc 3.1.1 @ /usr/local/bin/ (via rmarkdown) ## ## ─ Packages ─────────────────────────────────────────────────────────────────── ## package * version date (UTC) lib source ## bookdown 0.46 2025-12-05 [1] CRAN (R 4.3.2) ## bslib 0.6.1 2023-11-28 [1] RSPM (R 4.3.0) ## cachem 1.0.8 2023-05-01 [1] RSPM (R 4.3.0) ## cli 3.6.5 2025-04-23 [1] CRAN (R 4.3.2) ## devtools 2.4.5 2022-10-11 [1] RSPM (R 4.3.0) ## digest 0.6.34 2024-01-11 [1] RSPM (R 4.3.0) ## ellipsis 0.3.2 2021-04-29 [1] RSPM (R 4.3.0) ## evaluate 1.0.5 2025-08-27 [1] CRAN (R 4.3.2) ## fastmap 1.1.1 2023-02-24 [1] RSPM (R 4.3.0) ## fs 1.6.3 2023-07-20 [1] RSPM (R 4.3.0) ## glue 1.7.0 2024-01-09 [1] RSPM (R 4.3.0) ## htmltools 0.5.7 2023-11-03 [1] RSPM (R 4.3.0) ## htmlwidgets 1.6.4 2023-12-06 [1] RSPM (R 4.3.0) ## httpuv 1.6.14 2024-01-26 [1] RSPM (R 4.3.0) ## jquerylib 0.1.4 2021-04-26 [1] RSPM (R 4.3.0) ## jsonlite 2.0.0 2025-03-27 [1] CRAN (R 4.3.2) ## knitr 1.50 2025-03-16 [1] CRAN (R 4.3.2) ## later 1.3.2 2023-12-06 [1] RSPM (R 4.3.0) ## lifecycle 1.0.4 2023-11-07 [1] RSPM (R 4.3.0) ## magrittr 2.0.3 2022-03-30 [1] RSPM (R 4.3.0) ## memoise 2.0.1 2021-11-26 [1] RSPM (R 4.3.0) ## mime 0.12 2021-09-28 [1] RSPM (R 4.3.0) ## miniUI 0.1.1.1 2018-05-18 [1] RSPM (R 4.3.0) ## pkgbuild 1.4.3 2023-12-10 [1] RSPM (R 4.3.0) ## pkgload 1.4.1 2025-09-23 [1] CRAN (R 4.3.2) ## profvis 0.3.8 2023-05-02 [1] RSPM (R 4.3.0) ## promises 1.2.1 2023-08-10 [1] RSPM (R 4.3.0) ## purrr 1.0.2 2023-08-10 [1] RSPM (R 4.3.0) ## R6 2.6.1 2025-02-15 [1] CRAN (R 4.3.2) ## Rcpp 1.0.12 2024-01-09 [1] RSPM (R 4.3.0) ## remotes 2.4.2.1 2023-07-18 [1] RSPM (R 4.3.0) ## rlang 1.1.6 2025-04-11 [1] CRAN (R 4.3.2) ## rmarkdown 2.25 2023-09-18 [1] RSPM (R 4.3.0) ## sass 0.4.8 2023-12-06 [1] RSPM (R 4.3.0) ## sessioninfo 1.2.2 2021-12-06 [1] RSPM (R 4.3.0) ## shiny 1.8.0 2023-11-17 [1] RSPM (R 4.3.0) ## stringi 1.8.3 2023-12-11 [1] RSPM (R 4.3.0) ## stringr 1.5.1 2023-11-14 [1] RSPM (R 4.3.0) ## urlchecker 1.0.1 2021-11-30 [1] RSPM (R 4.3.0) ## usethis 2.2.3 2024-02-19 [1] RSPM (R 4.3.0) ## vctrs 0.6.5 2023-12-01 [1] RSPM (R 4.3.0) ## xfun 0.55 2025-12-16 [1] CRAN (R 4.3.2) ## xtable 1.8-4 2019-04-21 [1] RSPM (R 4.3.0) ## yaml 2.3.12 2025-12-10 [1] CRAN (R 4.3.2) ## ## [1] /usr/local/lib/R/site-library ## [2] /usr/local/lib/R/library ## ## ────────────────────────────────────────────────────────────────────────────── "],["references.html", "References", " References "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]] +[["clinical-data-management.html", "Chapter 5 Clinical Data Management 5.1 Learning Objectives 5.2 Clinical data handling tools 5.3 Privacy considerations for clinical data 5.4 Government regulators 5.5 Documentation 5.6 Conclusion", " Chapter 5 Clinical Data Management 5.1 Learning Objectives 5.2 Clinical data handling tools Clinical research relies heavily on the collection, processing, analysis, and management of data. Efficient and effective data handling is crucial to ensure the integrity, reliability, and validity of clinical trials and studies. Selecting the right tools for handling clinical data is a critical step in the research process. This section defines data handling in the context of clinical research and contrasts it with data quality, discusses the importance of privacy, provides a comprehensive guide on how to find suitable clinical data handling tools with a focus on open-source options, and emphasizes the importance of testing data handling methods using dummy data in the preliminary phases of a clinical trial. 5.2.1 Clinical data handling Clinical data handling involves the processes and tools used to collect, manage, store, process, and share clinical data in a way that maintains its integrity, security, and usability. It encompasses a range of activities, including: Data Collection: Gathering data from various sources such as case report forms (CRFs), electronic health records (EHRs), patient surveys, and laboratory results. Data Storage and Management: Safely storing data in databases or data warehouses and managing access to ensure that only authorized personnel can interact with the data. Data Cleaning and Processing: Ensuring data consistency, accuracy, and completeness by detecting and rectifying errors, missing values, and inconsistencies. Data Analysis: Using statistical and analytical tools to generate insights from the data that can inform clinical decisions and study outcomes. Data Sharing and Reporting: Providing access to data for collaborators, regulators, or stakeholders in a secure and controlled manner and generating reports that summarize findings. The goal of effective data handling is to ensure that data is accurate, reliable, and available when needed, while also protecting patient privacy and complying with regulatory standards. 5.2.1.1 Data quality vs data handling Data quality and data handling are closely related but distinct concepts in clinical research. Data Quality: Refers to the accuracy, completeness, consistency, and reliability of data. High-quality data is essential for producing valid and reliable research outcomes. It involves processes such as data validation, error checking, and quality control measures. Data Handling: Encompasses the broader scope of managing the data lifecycle, from collection and storage to processing, analysis, and sharing. While data quality is a component of data handling, the latter also involves aspects like data security, access management, and regulatory compliance. Both data quality and data handling are critical for ensuring the integrity of clinical research. Data handling tools must support high data quality standards through features like automated data validation checks, error reporting, and data cleaning functions. 5.2.2 Selecting clinical data handling tools Choosing the right tools for clinical data handling depends on several factors, including the study’s size, complexity, data types, regulatory requirements, and budget. Key considerations include: Compatibility and Integration: The tool should be compatible with existing data systems and workflows, allowing for seamless integration of data from multiple sources. Scalability: The tool should be able to handle the volume of data expected in the study and scale as the study progresses. User-Friendliness: A user-friendly interface can reduce the learning curve for researchers and data managers, improving efficiency and reducing errors. Regulatory Compliance: The tool should comply with relevant data privacy and security regulations to protect patient data. Support and Documentation: Adequate support, including user manuals, tutorials, and customer service, is essential for troubleshooting and maximizing the tool’s potential. Research teams should conduct thorough evaluations, including reading reviews, seeking recommendations, and conducting pilot tests, to identify the best tool for their specific needs. 5.2.2.1 Open source options Open-source tools provide a cost-effective and flexible alternative to proprietary software for handling clinical data. These tools are often developed and maintained by vibrant communities and can be customized to fit specific research needs. Advantages of open-source tools include: They are typically free to use They provide transparency by allowing users to review the codebase for security and compliance They can be modified to meet unique requirements They often have strong community support, which can be valuable for troubleshooting and enhancing functionalities. Examples of popular open-source data handling tools include: REDCap: A secure, web-based application for building and managing online surveys and databases, widely used in clinical research for data collection and management. OpenClinica: An open-source clinical trial software that supports data capture, management, and analysis, compliant with regulatory standards like FDA requirements for electronic data (FDA 21 CFR Part 11). KNIME: A data analytics, reporting, and integration platform that can be used for data cleaning, processing, and analysis, with extensive capabilities for machine learning and data visualization. R and Python: These programming languages offer powerful libraries and packages (such as the tidyverse for R, and pandas and NumPy for Python) that support a wide range of data handling and analysis tasks. Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform that offers some basic statistical and data visualization tools, and offers APIs to allow the real-time analytics in other platforms (R, C#, cURL, etc.). However, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued guidance on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR. While open-source tools are beneficial, it is essential to ensure they are secure, well-maintained, and compliant with relevant privacy and regulatory standards. When moving between multiple data handling tools, it is important to consider interoperability. As an example, REDCap and R play very nicely together via API tokens, but this may not be true of all electronic data capture systems or databases. The sharing, integration and redistribution of data between systems - whether internally or externally - needs to be carefully planned, tested and documented. 5.2.3 Testing data handling tools Before implementing data handling methods in an actual clinical trial, it is crucial to run tests using dummy data. This step ensures that the data handling process is robust, efficient, and free from errors across different conditions, without risking sensitive patient information. Creating Realistic Dummy Data: Dummy data should mimic the real data in terms of format, structure, and complexity. It should include various scenarios (e.g., missing data, outliers, data entry errors) to test the system’s error-handling capabilities. Simulating the Full Data Handling Workflow: The process should involve every stage of data handling, from data collection and entry to storage, cleaning, analysis, and reporting. This comprehensive simulation helps identify potential issues early, such as data loss, security vulnerabilities, or errors in data processing algorithms. Refining and Optimizing Data Handling Methods: Based on the findings from dummy data tests, researchers can refine their data handling protocols, adjust tools and settings, and optimize workflows to ensure smooth operations when real data is introduced. Testing with dummy data provides an additional layer of quality assurance and helps build confidence in the data handling process before the clinical trial begins. 5.2.4 Conclusion Effective clinical data handling is fundamental to the success of clinical trials and studies. Selecting the right tools requires careful consideration of privacy, regulatory compliance, open-source versus proprietary options, and the specific needs of the study. Understanding the difference between data quality and data handling is crucial, as both are essential for ensuring the validity and reliability of research outcomes. Running preliminary tests using dummy data is a critical step in validating data handling methods, ensuring that the chosen tools and processes are robust, secure, and efficient. By following these guidelines, researchers can enhance data management practices, protect patient privacy, and achieve meaningful and reliable clinical research outcomes. 5.3 Privacy considerations for clinical data Privacy is a fundamental consideration in clinical data management, given the sensitive nature of the information involved. Clinical data often contains personally identifiable information (PII) or personal health information (PHI), which must be protected to comply with privacy regulations such as the Health Insurance Portability and Accountability Act (HIPAA) in the United States, the General Data Protection Regulation (GDPR) in the European Union, and other regional laws. More information about PII and PHI can be found in this course about ethical data handling. Data Anonymization and De-identification: One of the primary methods to protect privacy is to anonymize or de-identify data, removing or encrypting any information that could directly or indirectly identify an individual. Privacy-Preserving Record Linkages: In scenarios where data from multiple sources need to be linked without compromising individual privacy, privacy-preserving record linkage techniques are essential. These methods enable the integration of datasets by matching records in a way that minimizes the risk of re-identification. Techniques such as secure multi-party computation, homomorphic encryption, and differential privacy can be employed to ensure that the linkage process itself does not expose sensitive information. Access Control and Encryption: Tools should support robust access controls, ensuring that only authorized users have access to the data. Encryption should be used for both data at rest and data in transit to prevent unauthorized access. Compliance and Auditing: Tools should facilitate compliance with regulatory standards and provide auditing capabilities to track data access and usage. When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research. 5.4 Government regulators Clinical data is governed by several different types of regulations. In this section, we will review some of the major regulatory frameworks and organizations. 5.4.1 Health Insurance Portability and Accountability Act (HIPAA) The Health Insurance Portability and Accountability Act (HIPAA), regulated by the U.S. Department of Health and Human Services (HHS), establishes national standards to protect individuals’ medical records and other personal health information. It applies to health plans, healthcare clearinghouses, and healthcare providers that conduct certain healthcare transactions electronically. The HIPAA Privacy Rule requires appropriate safeguards to protect the privacy of protected health information (PHI) and sets limits on the uses and disclosures of such information without patient authorization. 5.4.2 Food and Drug Administration (FDA) The U.S. Food and Drug Administration (FDA) regulates the safety, efficacy, and security of human and veterinary drugs, biological products, medical devices, food, cosmetics, and products that emit radiation. The FDA’s regulations ensure that clinical trials are conducted ethically and that data collected is reliable and accurate. This includes oversight of clinical trial protocols, informed consent, and reporting of adverse events. 5.4.3 General Data Protection Regulation (GDPR) The General Data Protection Regulation (GDPR), governed by the European Union (EU), is a comprehensive data protection law that governs the collection, processing, storage, and transfer of personal data within the EU. It aims to enhance individuals’ control over their personal data and simplify the regulatory environment for international business. GDPR applies to any organization that processes the personal data of EU residents, regardless of where the organization is based. Key provisions include the right to be forgotten, data portability, and mandatory breach notifications. 5.4.4 Honest brokers Honest brokers act as neutral intermediaries between the data source and researchers, typically regulated by Institutional Review Boards (IRBs) or equivalent bodies. They are responsible for de-identifying data to ensure that researchers cannot trace the data back to individual patients. Honest brokers must complete specific training, such as Collaborative Institutional Training Initiative (CITI) Research Ethics and HIPAA training, before accessing data. They play a crucial role in maintaining the confidentiality and integrity of clinical data. 5.5 Documentation In data management, several types of documentation are frequently encountered, each serving a specific purpose throughout the lifecycle of a research project. These documents are crucial for ensuring data integrity, regulatory compliance, and effective project management. 5.5.1 Data Management Plan A Data Management Plan (DMP) outlines how data will be handled during and after a research project. Typically created at the beginning of a project, it is stored in project documentation repositories or institutional databases. Often required by funding agencies, a DMP may include plans for testing with dummy data to ensure data integrity. The importance of a DMP lies in its role in planning and managing data throughout the project, ensuring that data is handled consistently and responsibly. 5.5.2 Statistical Analysis Plan The Statistical Analysis Plan (SAP) details the statistical methods and analyses to be performed on the collected data. Developed before data analysis begins, it is stored with project documentation or in electronic lab notebooks. Essential for clinical trials and other research requiring rigorous statistical analysis, the SAP may involve testing statistical methods on dummy data. The SAP is crucial for maintaining the integrity and reproducibility of statistical analyses. 5.5.3 Standard Operating Procedures Standard Operating Procedures (SOPs) provide detailed instructions on how to perform specific tasks or processes. Used throughout the project lifecycle, they are stored in organizational repositories or document management systems. SOPs are often required for regulatory compliance and may include procedures for testing with dummy data. The importance of SOPs lies in their ability to standardize processes, ensuring consistency and compliance with regulatory requirements. 5.5.4 Data Use Agreements Data Use Agreements (DUAs) define the terms and conditions for data sharing and use. Encountered before data is shared with external parties, they are stored in legal or administrative offices. DUAs are required when sharing data with external collaborators but are typically not directly related to dummy data. DUAs are important for protecting data privacy and ensuring that data is used appropriately. 5.5.5 Data Sharing Agreements Data Sharing Agreements (DSAs) specify the terms for sharing data between organizations. Encountered prior to data sharing, they are stored in legal or administrative offices. DSAs are necessary for formalizing data sharing arrangements and generally do not relate to dummy data. The importance of DSAs lies in their role in facilitating collaboration while protecting data integrity and compliance with legal requirements. 5.5.6 Documentation across the span of a project These documents are encountered at various stages of a project, from planning (DMP, SAP) to execution (SOP) and data sharing (DUA, DSA). They are typically stored in project documentation repositories, institutional databases, or document management systems, with access restricted to authorized personnel. Not all documents are required for every project; their necessity depends on the project’s scope, regulatory requirements, and institutional policies. Some documents, like the DMP, SAP, and SOP, may include provisions for testing with dummy data to ensure data integrity and validate processes. To ensure compliance with these documents, organizations should implement robust data governance frameworks that include regular audits, training programs, and clear policies and procedures. Standardized templates for these documents are often available from funding agencies, regulatory bodies, or institutional guidelines, helping to ensure consistency and compliance with best practices. It’s best to ask before starting any document to ensure the correct format is used. For further reading, consider the following resources: Guide to Clinical Data Management Procedures (GCDMP): GCDMP Books by Suzanne Prokscha: Writing and Managing SOPs for GCP and Practical Guide to Clinical Data Management Additionally, remember the insightful quote by Damian Conway: “Documentation is a love letter that you write to your future self.” This is an invaluable tidbit to keep in mind throughout the lifecycle of a study. 5.6 Conclusion "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]] From 48213c294a69b811bcd9f7506b61691d4b453691 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 13:16:47 -0700 Subject: [PATCH 04/12] Apply suggestions from code review Co-authored-by: Kate Isaac <41767733+kweav@users.noreply.github.com> --- 03-data_management.Rmd | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index 2f11adc..2ae2887 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -41,7 +41,7 @@ Both data quality and data handling are critical for ensuring the integrity of c ### Selecting clinical data handling tools -Choosing the right tools for clinical data handling depends on several factors, including the study's size, complexity, data types, regulatory requirements, and budget. Key considerations include: +Choosing the right tools for clinical data handling depends on several factors, including the study's size, complexity, data types, regulatory requirements, and budget. Clinical data studies will often utilize multiple tools (e.g., at least one for data collection and another for data analysis). Depending on what stage the study is in, researchers may not need to choose tools for all or certain aspects of the clinical data handling process, for example because data collection may have already occurred. Key considerations include: - **Compatibility and Integration**: The tool should be compatible with existing data systems and workflows, allowing for seamless integration of data from multiple sources. - **Scalability**: The tool should be able to handle the volume of data expected in the study and scale as the study progresses. @@ -69,11 +69,11 @@ Examples of popular open-source data handling tools include: - **KNIME**: A data analytics, reporting, and integration platform that can be used for data cleaning, processing, and analysis, with extensive capabilities for machine learning and data visualization. - **R and Python**: These programming languages offer powerful libraries and packages (such as the `tidyverse` for R, and `pandas` and `NumPy` for Python) that support a wide range of data handling and analysis tasks. -Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform that offers some basic statistical and data visualization tools, and offers APIs to allow the real-time analytics in other platforms (R, C#, cURL, etc.). However, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued [guidance](https://www.r-project.org/doc/R-FDA.pdf) on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR. +Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform that offers some basic statistical and data visualization tools, and offers APIs to allow for real-time analytics in other platforms (R, C#, cURL, etc.). However, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued [guidance](https://www.r-project.org/doc/R-FDA.pdf) on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR. While open-source tools are beneficial, it is essential to ensure they are secure, well-maintained, and compliant with relevant privacy and regulatory standards. -When moving between multiple data handling tools, it is important to consider interoperability. As an example, REDCap and R play very nicely together via API tokens, but this may not be true of all electronic data capture systems or databases. The sharing, integration and redistribution of data between systems - whether internally or externally - needs to be carefully planned, tested and documented. +When moving between multiple data handling tools, it is important to consider interoperability. As an example, REDCap and R play very nicely together via API tokens, but this may not be true of all electronic data capture systems or databases. The sharing, integration and redistribution of data between systems - whether internally or externally - needs to be carefully planned, tested, and documented. ### Testing data handling tools From 8dff38757d9f0efa9923b330b3acfe3ce582dd0f Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 13:31:51 -0700 Subject: [PATCH 05/12] add note about considering institutional support when choosing data handling tools --- 03-data_management.Rmd | 1 + 1 file changed, 1 insertion(+) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index 2ae2887..fb1ff2d 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -48,6 +48,7 @@ Choosing the right tools for clinical data handling depends on several factors, - **User-Friendliness**: A user-friendly interface can reduce the learning curve for researchers and data managers, improving efficiency and reducing errors. - **Regulatory Compliance**: The tool should comply with relevant data privacy and security regulations to protect patient data. - **Support and Documentation**: Adequate support, including user manuals, tutorials, and customer service, is essential for troubleshooting and maximizing the tool's potential. +- **Institutional Support**: Your institution may have policies, regulations, or recommendations about which tools can be used—particularly when working with clinical data. Institutions often provide licenses or access to specific tools (for example, an institutional REDCap consortium membership), as well as varying levels of support through scientific staff, data offices, or training programs. Research teams should conduct thorough evaluations, including reading reviews, seeking recommendations, and conducting pilot tests, to identify the best tool for their specific needs. From af328f0d49061c74b508ba47bfe83e624a1e64c2 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 13:40:51 -0700 Subject: [PATCH 06/12] add note about how mainstream tools can be safe options, as opposed to newer or less common tools --- 03-data_management.Rmd | 1 + 1 file changed, 1 insertion(+) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index fb1ff2d..c56f95c 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -49,6 +49,7 @@ Choosing the right tools for clinical data handling depends on several factors, - **Regulatory Compliance**: The tool should comply with relevant data privacy and security regulations to protect patient data. - **Support and Documentation**: Adequate support, including user manuals, tutorials, and customer service, is essential for troubleshooting and maximizing the tool's potential. - **Institutional Support**: Your institution may have policies, regulations, or recommendations about which tools can be used—particularly when working with clinical data. Institutions often provide licenses or access to specific tools (for example, an institutional REDCap consortium membership), as well as varying levels of support through scientific staff, data offices, or training programs. +- **Established Practices in the Field**: Choosing tools that are widely used and accepted in your field is often a safe and effective option. Although newer or less common tools may offer advantages, using them may require additional justification to explain why they are suitable for your specific use case. Research teams should conduct thorough evaluations, including reading reviews, seeking recommendations, and conducting pilot tests, to identify the best tool for their specific needs. From e4b4dc1ac42309aab1586d63e45a0db82c6a8234 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 13:44:20 -0700 Subject: [PATCH 07/12] add note about redcap's ability for data quality checks, and that it can do basic repots but not advanced statistical modeling or analyses --- 03-data_management.Rmd | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index c56f95c..ff48995 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -71,7 +71,7 @@ Examples of popular open-source data handling tools include: - **KNIME**: A data analytics, reporting, and integration platform that can be used for data cleaning, processing, and analysis, with extensive capabilities for machine learning and data visualization. - **R and Python**: These programming languages offer powerful libraries and packages (such as the `tidyverse` for R, and `pandas` and `NumPy` for Python) that support a wide range of data handling and analysis tasks. -Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform that offers some basic statistical and data visualization tools, and offers APIs to allow for real-time analytics in other platforms (R, C#, cURL, etc.). However, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued [guidance](https://www.r-project.org/doc/R-FDA.pdf) on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR. +Each of these tools carries their own idiosyncratic pros and cons. For example, KNIME is a code-free analytical platform that may be ideal for students or new analysts to get comfortable with data processing and management. However, KNIME is not easily scalable, so it may not be ideal for large or multi-site projects. REDCap is a highly scalable and customizable survey and data collection platform, and has robust capability for data quality checks. While it offers some basic statistical and data visualization tools, and has APIs to allow for real-time analytics in other platforms (R, C#, cURL, etc.), it is not capable of advanced statistical modeling or more sophisticated analyses that are possible with R and Python. Additionally, REDCap licensing limits it to non-profit institutions (and other potentially limiting requirements), so this should be investigated before plans are made to use the environment. REDCap and R both allow for regulatory compliance. The FDA has issued [guidance](https://www.r-project.org/doc/R-FDA.pdf) on using R for clinical trials. REDCap can be configured to support compliance with the FDA 21 CFR Part 11, HIPAA, and GDPR. While open-source tools are beneficial, it is essential to ensure they are secure, well-maintained, and compliant with relevant privacy and regulatory standards. From 04ee3afbdd0a348a97688c6b0e7d74cf227bd600 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 14:03:14 -0700 Subject: [PATCH 08/12] moving around note on privacy features of data tools, before giving specific examples --- 03-data_management.Rmd | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index ff48995..ab2c5c4 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -87,6 +87,13 @@ Before implementing data handling methods in an actual clinical trial, it is cru Testing with dummy data provides an additional layer of quality assurance and helps build confidence in the data handling process before the clinical trial begins. +#### How to Generate Dummy Data + +There are several ways to simulate data to use for testing. + +- Write a script that uses a probabilistic modeling framework to create a dataset with values that mimic what you expect to find. You can also remove data or add errors that might occur in your real data collection process +- If you are using a case report form to collect data, you could answer questions several times to create a dataset, and then continue through the data handling pipeline + ### Conclusion Effective clinical data handling is fundamental to the success of clinical trials and studies. Selecting the right tools requires careful consideration of privacy, regulatory compliance, open-source versus proprietary options, and the specific needs of the study. Understanding the difference between data quality and data handling is crucial, as both are essential for ensuring the validity and reliability of research outcomes. Running preliminary tests using dummy data is a critical step in validating data handling methods, ensuring that the chosen tools and processes are robust, secure, and efficient. By following these guidelines, researchers can enhance data management practices, protect patient privacy, and achieve meaningful and reliable clinical research outcomes. @@ -95,13 +102,13 @@ Effective clinical data handling is fundamental to the success of clinical trial Privacy is a fundamental consideration in clinical data management, given the sensitive nature of the information involved. Clinical data often contains personally identifiable information (PII) or personal health information (PHI), which must be protected to comply with privacy regulations such as the [Health Insurance Portability and Accountability Act](https://www.hhs.gov/hipaa/for-professionals/privacy/laws-regulations/index.html) (HIPAA) in the United States, the [General Data Protection Regulation](https://gdpr.eu/what-is-gdpr/) (GDPR) in the European Union, and other regional laws. More information about PII and PHI can be found in [this course](https://hutchdatascience.org/Ethical_Data_Handling_for_Cancer_Research/data-privacy.html) about ethical data handling. +When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research. + - **Data Anonymization and De-identification**: One of the primary methods to protect privacy is to anonymize or de-identify data, removing or encrypting any information that could directly or indirectly identify an individual. - **Privacy-Preserving Record Linkages**: In scenarios where data from multiple sources need to be linked without compromising individual privacy, privacy-preserving record linkage techniques are essential. These methods enable the integration of datasets by matching records in a way that minimizes the risk of re-identification. Techniques such as secure multi-party computation, homomorphic encryption, and differential privacy can be employed to ensure that the linkage process itself does not expose sensitive information. - **Access Control and Encryption**: Tools should support robust access controls, ensuring that only authorized users have access to the data. Encryption should be used for both data at rest and data in transit to prevent unauthorized access. - **Compliance and Auditing**: Tools should facilitate compliance with regulatory standards and provide auditing capabilities to track data access and usage. -When selecting data handling tools, it is crucial to prioritize those that offer comprehensive privacy features to safeguard sensitive information and maintain public trust in clinical research. - ## Government regulators Clinical data is governed by several different types of regulations. In this section, we will review some of the major regulatory frameworks and organizations. From 22e992dcc909cbaeee6690ff2ec5b5f5abc12d18 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 14:11:00 -0700 Subject: [PATCH 09/12] add subsection on how to generate dummy data for testing --- 03-data_management.Rmd | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index ab2c5c4..031ba60 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -91,8 +91,9 @@ Testing with dummy data provides an additional layer of quality assurance and he There are several ways to simulate data to use for testing. -- Write a script that uses a probabilistic modeling framework to create a dataset with values that mimic what you expect to find. You can also remove data or add errors that might occur in your real data collection process +- Write a script that uses a probabilistic modeling framework to create a dataset with values that mimic what you expect to find. There are also R packages that can help with this task, such as [`simstudy`](https://cran.r-project.org/web/packages/simstudy/index.html). After you simulate data, can also remove data or add errors that might occur in your real data collection process - If you are using a case report form to collect data, you could answer questions several times to create a dataset, and then continue through the data handling pipeline +- In some scenarios, you could start with a real dataset that is publicly available and de-identified, and then directly use that data or sample from that data, adding missing data and errors that might occur in your data generation process ### Conclusion From 9bb8735cb270ce83b4efd2523e926eda412b023f Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 14:12:14 -0700 Subject: [PATCH 10/12] add note about refining surveys/case report forms from testing --- 03-data_management.Rmd | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/03-data_management.Rmd b/03-data_management.Rmd index 031ba60..821a510 100644 --- a/03-data_management.Rmd +++ b/03-data_management.Rmd @@ -83,7 +83,7 @@ Before implementing data handling methods in an actual clinical trial, it is cru - **Creating Realistic Dummy Data**: Dummy data should mimic the real data in terms of format, structure, and complexity. It should include various scenarios (e.g., missing data, outliers, data entry errors) to test the system's error-handling capabilities. - **Simulating the Full Data Handling Workflow**: The process should involve every stage of data handling, from data collection and entry to storage, cleaning, analysis, and reporting. This comprehensive simulation helps identify potential issues early, such as data loss, security vulnerabilities, or errors in data processing algorithms. -- **Refining and Optimizing Data Handling Methods**: Based on the findings from dummy data tests, researchers can refine their data handling protocols, adjust tools and settings, and optimize workflows to ensure smooth operations when real data is introduced. +- **Refining and Optimizing Data Handling Methods**: Based on the findings from dummy data tests, researchers can refine their data handling protocols, such as their survey instruments or case report forms, adjust tools and settings, and optimize workflows to ensure smooth operations when real data is introduced. Testing with dummy data provides an additional layer of quality assurance and helps build confidence in the data handling process before the clinical trial begins. From a7ed3d7dd2efeb0218bbe70622b5c1c21098bb17 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Thu, 18 Jun 2026 14:29:28 -0700 Subject: [PATCH 11/12] add .gitignore back in, add docs and .Rproj.user folder --- .gitignore | 12 ++++++++++++ 1 file changed, 12 insertions(+) create mode 100644 .gitignore diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000..87ea9dc --- /dev/null +++ b/.gitignore @@ -0,0 +1,12 @@ +# ignore nonsense files +.DS_Store +.Rhistory +.local +.rstudio +.bash_history +spell_check_results.tsv +.RData +.httr-oauth +docker/git_token.txt +docs/* +.RProj.user/ From 81015dfe46ab19725b7c21d50b262be450f14a02 Mon Sep 17 00:00:00 2001 From: Sarah Teichman Date: Mon, 22 Jun 2026 10:53:29 -0700 Subject: [PATCH 12/12] add terms to dictionary from spellcheck --- resources/dictionary.txt | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) diff --git a/resources/dictionary.txt b/resources/dictionary.txt index 5d61f70..c96aa77 100644 --- a/resources/dictionary.txt +++ b/resources/dictionary.txt @@ -1,4 +1,7 @@ al +analytics +Anonymization +anonymize ASR biomarker biomarkers @@ -9,37 +12,58 @@ bookdown Bookdown CDS centric +CFR chatbots +CITI +codebase Coursera Collaboratory colorectal comorbidities Comorbidities CPT +CRF CRFs CUIs +cURL +customizable customizations DaSL data's de +DMP +DSA +DSAs +DUA +DUAs et EHR faq +FDA's FHIR GARDE +GCDMP +GCP +GDPR generalizability generalizable HAQ HD +HIPPA +homomorphic https ICD Immunohistochemical +individuals' +IRBs ITCR itcrtraining ITN JK Kelleher +KNIME Leanpub +lifecycle LLM LLMs LOINC @@ -53,6 +77,7 @@ NDC NIH's NLP OMOP +OpenClinica ontologies participants' perscribed @@ -60,23 +85,30 @@ Permanente personalization Ph PHM +PII polypharmacy pre +Prokscha PROMs QALY QALYs +REDCap representativeness +reproducibility Robbertz RMarkdown RWD RWE RxNorm +Scalability +scalable SAEs sexualized Shriver SNOMED socio SOGI +SOPs telehealth templated Tolerability