MedNav is a conversational AI system that guides users from symptoms to actionable care decisions, including provider recommendations and emergency escalation.
Built for HooHacks 2026.
Created by Eric Xie and Siwen Liao
- Multi-turn symptom reasoning
- Adaptive follow-up questioning
- Evidence-based guidance
- Provider recommendation with review-based justification
- Location-aware care navigation (maps)
- Emergency detection and escalation
- Visit preparation summaries
The system is built as a multi-agent conversational pipeline:
User Input
→ Intake Agent
→ Retrieval Agent
→ Moderator (decision-making)
→ Response Formatter
→ Care Finder
→ Visit Summary
All interactions are handled in a single conversational thread.
pip install -r requirements.txt
Create a .env file:
GOOGLE_API_KEY=your_api_key_here
uvicorn main:app --reload
Open in browser:
http://localhost:8000/health
MedNav goes beyond symptom checking by:
- Handling uncertainty through targeted questioning
- Providing explainable recommendations (reviews + reasoning)
- Connecting users directly to care
- Reducing delays in emergency situations
- FastAPI (backend API)
- Pydantic (structured schemas)
- Google Generative AI (google.genai)
- Python
- This project is designed as a chat-first system, not a multi-page workflow.
- All interactions happen in a single conversational thread.
- The backend maintains context across turns.