A complete AI-powered medical transcription application with a FastAPI backend and Next.js frontend. This platform processes audio files of medical dictations and returns structured clinical data including transcriptions, SOAP notes, billing codes, and prescriptions.

• 🎙️ Real-time Audio Streaming: Live transcription with WebSocket communication
• 📁 File Upload: Traditional audio file processing
• 🏥 Medical Intelligence: Automatic SOAP note generation
• 💰 Billing Integration: Real-time ICD-10 code lookup
• 💊 Prescription Management: Structured medication extraction
• 🔬 Lab Orders: Automated test ordering
• ⚡ Professional UI: Medical-themed, responsive interface

pip install -r requirements.txt

uvicorn main:app --host 0.0.0.0 --port 8000 --reload

The API will be available at http://localhost:8000

npm install

npm run dev

The web app will be available at http://localhost:3000

Option 1: Use the Web Interface

• Go to http://localhost:3000
• Choose "Real-time Streaming" or "File Upload" mode
• Test the live transcription or upload an audio file

Option 2: Test APIs Directly

Traditional HTTP endpoint:

python test_api.py

WebSocket streaming endpoint:

python test_websocket.py

Real-time audio streaming endpoint for live transcription

Connection: ws://localhost:8000/ws/process-visit

Message Flow:

1. Client → Server: Binary audio chunks (bytes)
2. Server → Client: Partial transcription updates

   {
     "type": "partial_transcript",
     "text": "Patient is a 34-year-old...",
     "is_final": false
   }

3. Client → Server: "END_OF_STREAM" (text message)
4. Server → Client: Final structured result

   {
     "type": "final_result",
     "data": {
       "transcription": "Patient is a 34-year-old male presenting with...",
       "soap_note": "SUBJECTIVE:\n34-year-old male presents with...",
       "diagnosis": "streptococcal pharyngitis",
       "billing_code": {
         "code": "J02.0",
         "description": "streptococcal pharyngitis"
       },
       "prescriptions": [...],
       "lab_orders": [...]
     }
   }

Traditional file upload endpoint (fallback)

Accepts an audio file upload and returns structured medical data.

Request:

• Method: POST
• Content-Type: multipart/form-data
• Body: Audio file (any format)

Response: Same structure as WebSocket final result data

Health check endpoint that returns a simple status message.

├── main.py              # Main FastAPI application
├── requirements.txt     # Python dependencies
├── data/
│   └── icd10_codes.csv # ICD-10 billing codes database
├── test_api.py          # HTTP API testing script
├── test_websocket.py    # WebSocket streaming test script
└── README.md           # This file

• Speech-to-Text: Returns hardcoded transcription
• Entity Extraction: Extracts diagnosis, medication, dosage, etc.
• SOAP Note Generation: Creates structured clinical notes

• Billing Code Lookup: Real CSV-based ICD-10 code lookup
• File Upload Handling: Proper multipart/form-data processing
• WebSocket Streaming: Real-time bidirectional communication
• Concurrent Processing: Async audio streaming with concurrent transcription
• CORS Support: Enabled for frontend development

This backend is ready for deployment to services like:

• Render (recommended for hackathons)
• Google Cloud Run
• Heroku
• AWS Lambda (with minor modifications)

To integrate real AI services, replace the mock functions in main.py:

1. transcribe_audio_mock() → OpenAI Whisper API
2. extract_entities_mock() → LLM API call with NER prompt
3. generate_soap_note_mock() → LLM API call with SOAP generation prompt

• Response Time: ~500ms (with mocks)
• File Upload: Supports any audio format
• Billing Lookup: Fast in-memory CSV lookup
• CORS: Configured for frontend development

• File upload validation
• Graceful CSV loading with fallbacks
• Billing code lookup with default responses
• Proper HTTP status codes