🩺 DoctorAI — Advanced AI Cardiologist (WHO + AHA Guided)

DoctorAI is a clinical‑grade cardiology support system powered by LLMs (Ollama / Llama 3), WHO 2025 guidelines, medical symptom mapping, OCR analysis, and intelligent emergency classification.

This is NOT a simple chatbot — it is a full medical AI framework with:

• Emergency detection (AHA/WHO compliant)
• Local medical database fusion
• WHO cardiology data integration
• Symptom-to-disease reasoning
• OCR support for ECG / reports
• Session memory + anonymized logging
• Streaming responses (ChatGPT-style)

It is designed for educational + research purposes only.

DoctorAI classifies user symptoms into:

• LEVEL 1 — EMERGENCY → “Call 108 / go to hospital NOW”
• LEVEL 2 — URGENT → Consult doctor within 24 hours
• LEVEL 3 — NON-URGENT → General guidance

Uses AHA 2025 + ESC 2024 approved symptom patterns.

Maps symptoms to possible diseases using symptom_disease_map.json.

Example:

'chest pain' → possible: Angina, MI, GERD

DoctorAI pulls data from:

• cardio_faq.json (common questions)
• cardio_tips.json (verified health tips)
• clinics.json (local cardiology centers)
• who_cardiology_data.json (WHO medical data)

All relevant information is automatically merged into the LLM prompt.

If user uploads files like:

• ECG
• Cholesterol report
• Blood test
• Medical scan

DoctorAI extracts text using Tesseract OCR and includes it in analysis.

Responses are streamed chunk-by-chunk for smooth real-time conversation.

WHO 2025 AI Ethics built-in:

• No diagnosis
• No prescriptions
• Transparency
• User consent system
• Anonymized audit logs (HIPAA-like)

Every response ends with a mandatory disclaimer.

• Remembers last 5 user questions (only if consented)
• Full anonymized logs
• Smart caching for repeated queries

DoctorAI/
│── app.py                  # Main Flask server
│── uploads/                # User uploaded medical files
│── local_db/
│     ├── cardio_faq.json
│     ├── cardio_tips.json
│     ├── clinics.json
│     ├── symptom_disease_map.json
│     ├── ai_cache.json
│     └── session_memory.json
│
├── data_sources/
│     └── who_cardiology_data.json
│
└── index.html              # Optional frontend UI

1. User sends message → /chat endpoint

2. System checks:

   • Local DB matches
   • WHO data
   • Symptom patterns
   • Emergency triggers

3. Builds a smart medical prompt

4. Passes to Ollama → streams response

5. Adds ethics disclaimer

6. Logs anonymized query

pip install flask flask-cors requests fuzzywuzzy python-Levenshtein pytesseract pillow

• Windows: https://github.com/UB-Mannheim/tesseract/wiki
• Linux: sudo apt install tesseract-ocr

ollama pull llama3:8b
ollama serve

python app.py

Server runs at:

http://localhost:5000

Chat with DoctorAI.

{
  "message": "I have chest pain radiating to my left arm"
}

Upload ECG / medical reports.

Health check.

DoctorAI is NOT a doctor. It provides educational guidance only. Always consult a certified cardiologist.

This is not just an LLM frontend. It is a complete medical intelligence pipeline:

• Risk classification
• Symptom inference
• WHO-level knowledge
• Real-time triage
• OCR medical analysis
• AI context merging

This puts the project in research + startup level domain.

Sanu Sharma — AI Engineer & Innovator

If you use this project, give credit to the original creator.

• Email: sanusharma0045.v@gmail.com