Transform clinical conversations into CMS-1500 claim forms with AI-powered automation.
MedRelay is a hackathon project that addresses the administrative burden of medical billing by automatically generating CMS-1500 claim forms from doctor-patient conversation transcripts. The system uses Google's Gemini AI to extract patient information, diagnoses, and procedures, then generates a properly formatted PDF claim form.
- AI-Powered Extraction: Automatically extracts patient details, diagnoses, and procedures from clinical transcripts
- Medical Code Lookup: Converts plain English medical terms to official ICD-10 and CPT codes
- PDF Generation: Creates properly formatted CMS-1500 claim forms
- Simple Web Interface: Easy-to-use frontend for transcript input and PDF download
- Real-time Processing: Fast AI processing with status indicators
- Backend: Python Flask
- AI: Google Gemini 2.0 Flash
- PDF Generation: fpdf2
- Frontend: HTML, JavaScript, Tailwind CSS
- Medical Coding: Custom lookup service
- Clone the repository:
git clone <repository-url>
cd scribemd- Install dependencies:
pip install -r requirements.txt- Set up environment variables:
# Create a .env file
echo "GEMINI_API_KEY=your_api_key_here" > .env- Run the application:
python app.py- Open your browser and navigate to
http://localhost:5000
- Input Transcript: Paste your doctor-patient conversation into the text area
- Generate Form: Click "Generate Claim Form" to process with AI
- Download PDF: Once processing is complete, download your CMS-1500 form
Doctor: Good morning, Sarah. What brings you in today?
Patient: I've been having a sore throat and fever for the past few days.
Doctor: Let me examine you. I can see some redness in your throat. I'll do a rapid strep test to check for strep throat.
Patient: Okay, that sounds good.
Doctor: The test came back positive for strep throat. I'll prescribe you some antibiotics and recommend rest and fluids.
Patient: Thank you, doctor.
GET /- Main web interfacePOST /generate-claim- Generate CMS-1500 form from transcriptGET /health- Health check endpoint
scribemd/
├── app.py # Main Flask application
├── pdf_generator.py # PDF generation service
├── code_lookup.py # Medical code lookup service
├── requirements.txt # Python dependencies
├── templates/
│ └── index.html # Frontend HTML page
└── static/ # Static assets (if needed)
The system currently supports lookup for common:
- ICD-10 Diagnosis Codes: Strep throat, fever, hypertension, diabetes, etc.
- CPT Procedure Codes: Office visits, rapid strep tests, injections, etc.
- This is a hackathon prototype and not intended for production use
- Medical code lookup is simplified and may not cover all cases
- PDF formatting is basic and may need refinement for specific requirements
- No validation of medical codes or claim accuracy
- Integration with real medical coding APIs
- Support for additional claim form types
- Enhanced PDF formatting and validation
- Batch processing capabilities
- Integration with EHR systems
This project is created for the HopHacks hackathon and is not intended for commercial use.