Stroke-Risk-Analytics

---

📌 Introduction

This project develops a modular Python-based analytics system to study the risk factors of stroke incidence using anonymized patient health records.
The system focuses on behavioural and clinical predictors such as hypertension, smoking status, and sleep patterns, aiming to support public health insights, clinical audits, and educational applications.

The motivation is to design a framework that is interpretable, auditable, and ethically aligned with healthcare data practices, while enabling scalable analysis for both descriptive and comparative studies.

---

📊 Dataset

• Source: Anonymized patient health records
• Features:
  • Stroke outcome (Yes/No)
  • Hypertension (binary)
  • Smoking status (categorical)
  • Sleep hours (numeric)
  • Diabetes, residence type, and comorbidities
• Data Challenges:
  • Missing and inconsistent values (esp. sleep & smoking)
  • Non-standardized categorical labels
  • Required preprocessing for cleaning and encoding

Preprocessing ensured uniformity and validity of the dataset for epidemiological analysis.

---

⚙️ Methodology

The methodology follows a modular, CLI-based analytical approach implemented in Python.

1. System Architecture

• dataset_module: Data ingestion, validation, and cleaning
• query_module: Analytical queries for risk factor relationships
• bimodule: Core logic combining results and extensions
• UI (CLI): User-friendly menu-driven interface
• Export Engine: Saves query outputs into CSV files

2. Analytical Queries

• Stroke prevalence among hypertensive smokers
• Cross-tabulation of residence type vs. stroke incidence
• Average sleep hours comparison between stroke-positive and negative groups
• Descriptive statistics (distribution, central tendency, subgroup breakdowns)

3. User Interface

• Command-Line Interface (CLI) with color-coded menus (via colorama)
• Designed for accessibility in low-resource environments
• Supports exporting results to CSV for audits and reproducibility

---

📈 Results

Key analytical findings:

• Hypertension + Smoking: Patients with both risk factors showed significantly higher stroke prevalence, consistent with medical literature.
• Sleep Patterns: Stroke-positive patients displayed slightly reduced sleep hours on average, aligning with recent clinical studies on sleep irregularity and vascular health.
• Descriptive Insights: Stratification by comorbidities and demographics revealed subgroup-specific trends valuable for targeted interventions.

These results demonstrate the potential of modular, transparent analytics in identifying actionable patterns in health datasets.

---

⚠️ Challenges & Limitations

• Data Quality Issues: Missing and inconsistent entries required extensive cleaning.
• Scope: Current design limited to descriptive and rule-based queries (no ML yet).
• Interface: CLI-only, may not appeal to all users compared to GUI dashboards.

---

🔮 Future Work

• Integration of machine learning models (logistic regression, random forest) for predictive risk scoring.
• Longitudinal analysis to track patients over time.
• Development of a GUI-based interface for broader adoption in clinical settings.

---

📚 References

• Feigin et al. (2025). World Stroke Organization Global Stroke Fact Sheet 2025.
• Zhou et al. (2025). The Stroke Burden in China and Its Long-Term Trends.
• Saeed et al. (2025). Machine learning-based stroke prediction.
• Jung et al. (2025). User-Centered Explanation Interfaces for AI in Healthcare.
• NIH StrokeNet (2025). Sleep Duration and Stroke Risk.

---