hatchmoment. scored by care · not by stars

TraceDoc

Local RAG‑powered clinical assistant with Streamlit UI

The project provides clinicians a quick, evidence‑based answer tool by retrieving from local guideline, literature, and case databases, then passing the results to an LLM. It ships a Streamlit front‑end and a FastAPI/uvicorn backend, uses SQLite for persistence, and includes scripts to build the knowledge base and smoke‑test the API. It is aimed at medical professionals or developers needing an offline, multi‑layered RAG system. Compared to generic chat bots, it offers built‑in three‑tier evidence display and easy Windows deployment without external services.

View on GitHub →

Wanjun-hd/TraceDoc