Multi-Agent-Medical-Assistant

Why we included this project

Healthcare teams wondering what GenAI assistants can realistically do in a sensitive domain will find a working reference here. The project coordinates several specialized agents for symptom triage, retrieval-augmented answers over clinical documents, and medical imaging analysis, wired together with LangGraph and exposed through a FastAPI service with a conversational chat interface. Its real strength is how it assembles production components instead of abstracting them: Qdrant handles hybrid vector search, Docling pulls text, tables, and images out of PDFs, a cross-encoder reranks retrieved chunks, and responses link back to the source documents. Clinicians and researchers evaluating open-source building blocks in medicine will appreciate that diagnosis conversations, image analysis, and live web research stay separate stages you can inspect rather than one opaque model call. Patients can try the demo, but treat this as an educational blueprint for agentic healthcare applications, not certified diagnostic software.