CT071-3-3-DDAC (DESIGNING & DEVELOPING CLOUD APPLICATIONS)

(Healthcare Industry) MedPal - AI-Powered Intelligent Medical Document Query Tool

Group 23

Name	TP Number
TEE HAO BIN	TP074435
LEE DE HONG	TP074501
MUHAMMAD ABID MAULANA DYSHAFALAH	TP069456
WEE ZHANG XUE	TP063516

Executive Summary

Healthcare professionals in Malaysia struggle to efficiently access critical information from vast, unstructured medical PDFs. Manual searches are time-consuming, error-prone, and limit clinical insights, impacting patient care. Our proposed Al query tool enables natural language smart search across all medical PDF documents via an intuitive chat-like interface.

Problem Background

1. Current Challenges in Malaysian Healthcare Doctors and administrative staff waste excessive time manually searching digital and physical archives for specific patient records or clinical guidelines stored as PDFs. Critical patient data and medical knowledge are often buried within unstructured PDF documents, hindering quick cross-referencing. Relying on keyword searches often leads to overlooked details, impacting accuracy. Without intelligent search, identifying trends or gaining deeper insights from collective document knowledge is challenging.

2. The Complexity of Medical Document Querying Medical documents come in diverse formats (scanned, legacy, digital) with complex terminology, abbreviations, and varying layouts. Traditional search methods struggle with unstructured text, especially when information is embedded within tables or graphics, making accurate retrieval difficult given the immense document volume.

Problem Definition

Create an Al-powered query tool that can intelligently search, extract, and synthesize information from all medical PDF documents within a healthcare institution. The solution should provide a "Gemini-like" chat interface, allowing healthcare professionals to ask natural language questions and receive precise, contextually relevant answers, effectively transforming unstructured documents into an accessible knowledge base.

Tech Stack

• Frontend - NextJS
• Backend - .NET
• Cloud - AWS

Project Structure

• frontend/ - NextJS frontend application
• backend/ - .NET backend application
• infrastructure/ - AWS infrastructure as code

Getting Started

Prerequisites

Backend Requirements:

• .NET 8.0 SDK ONLY
• Visual Studio 2022 or Visual Studio Code (recommended)
• AWS CLI configured with appropriate permissions

Frontend Requirements:

• Node.js (version 18 or higher)
• npm or yarn package manager
• TypeScript support

AWS Services Required:

• AWS Account with access to:
  • AWS Cognito (User authentication)
  • Amazon S3 (File storage)
  • Amazon DynamoDB (Database)
  • AWS Textract (Document processing)
  • AWS Lambda (Serverless functions)
  • AWS Elastic Beanstalk (Application hosting)

Development Tools:

• Git for version control
• Postman or similar API testing tool (optional)
• AWS Management Console access

Installation & Setup

1. Clone the repository

   git clone https://github.com/TeeHaoBin/DDAC_Assignment.git
   cd DDAC_Assignment

2. Install dependencies

   cd backend
   dotnet restore
   
   cd frontend
   npm install

3. Set up environment variables

   Frontend Environment:

   • Run cd frontend/ command
   • Copy .env.example to .env.local
   • Local Development: Configure NEXT_PUBLIC_API_BASE_URL to http://localhost:5149
   • Production: Configure NEXT_PUBLIC_API_BASE_URL to your domain URL deployed on AWS (e.g., https://your-api-domain.com)

   Backend Environment:

   • Configure AWS settings in backend/appsettings.json and backend/appsettings.Development.json
   • Set up AWS Cognito, S3, and DynamoDB connection strings

4. Run the development server

   cd backend
   dotnet run
   
   cd frontend
   npm run dev

5. Access the application

   • Open http://localhost:3000 in your browser
   • Main chat interface is available at the root path /