MedTwin

Your Medical Digital Twin -- A secure platform that gives patients full ownership of their medical records through FHIR-based data aggregation, AI-powered document processing, and portable QR code exports.

Built for the Harvard Health-Tech Hackathon 2026.

What It Does

MedTwin connects to hospital FHIR servers and consolidates a patient's scattered medical data into a single, patient-controlled profile. Patients can:

• View their complete medical history (conditions, medications, allergies, vitals, encounters)
• Upload medical documents (PDFs, FHIR bundles, HL7v2, C-CDA) that get automatically parsed and stored as structured FHIR resources
• Chat with an AI assistant that understands their medical context
• Export a portable PDF summary with a scannable QR code for easy sharing with new providers

Architecture

med-twin-app/          React + Vite frontend
app/                   FastAPI backend
  api/                 REST endpoints (patient, ingest, transfer, chat)
  services/            Core logic (FHIR client, PDF extraction, PDF generation, QR codes)
  models/              Pydantic schemas

Component	Technology
Frontend	React 18, Vite
Backend	FastAPI, Uvicorn
Health Data	Azure FHIR R4 Server
AI (Document Parsing)	GPT-4o via GitHub Models
AI (Chat)	GPT-4o via GitHub Models
PDF Generation	fpdf2
QR Codes	Nayuki QR Code Generator
Auth	API key + JWT transfer tokens

Getting Started

Prerequisites

• Python 3.12+
• Node.js 18+
• A GitHub PAT with Models (read) permission (free GPT-4o access)
• Azure FHIR server credentials

Setup

1. Clone the repo

   git clone https://github.com/LouayYa/MedTwin.git
   cd MedTwin

2. Configure environment

   cp .env.example .env
   # Edit .env with your credentials

3. Start the backend

   pip install -r requirements.txt
   uvicorn app.main:app --reload --port 8000

4. Start the frontend

   cd med-twin-app
   npm install
   npm run dev

5. Open http://localhost:5173 in your browser.

How It Works

Document Ingestion

Upload any medical document (PDF, FHIR JSON, HL7v2, C-CDA). PDFs are parsed by GPT-4o into structured FHIR R4 resources, which are validated and pushed to the Azure FHIR server.

QR Code Export

Generate a portable patient summary: FHIR data is retrieved, summarized by GPT-4o into a clean PDF, and a QR code is generated for easy sharing. The PDF can be downloaded directly from the browser.

AI Chat

Ask questions about your medical history. The AI assistant has access to your FHIR records and provides contextual, patient-friendly answers.

API Endpoints

Method	Endpoint	Description
GET	/api/v1/patients/lookup?emirates_id=	Find patient by Emirates ID
GET	/api/v1/patients/{id}/summary	Full medical summary
GET	/api/v1/patients/{id}/records	Import history
GET	/api/v1/patients/{id}/export	FHIR bundle export
POST	/api/v1/records/upload	Upload medical document
POST	/api/v1/transfer/export	Generate PDF + QR code
GET	/api/v1/transfer/pdf?token=	Download PDF via transfer token
POST	/api/claude	AI chat endpoint

Team

Built with care at the Harvard Health-Tech Hackathon 2026.

License

MIT