Inspiration Doctors spend a lot of time on documentation and administrative work instead of patient care. Hospitals also struggle with clinical decision support and monitoring patients after discharge, which can lead to complications. Doctorlib aims to help clinicians by acting as an AI clinical copilot that assists with analysis, documentation, and monitoring.
What it does Doctorlib is an AI-powered clinical intelligence platform that supports doctors during patient care.
It can:
Extract symptoms from clinical notes Analyze lab results and detect critical values Calculate clinical risk scores (MEWS / NEWS2) Generate discharge summaries and SOAP notes Monitor patients after discharge Predict risk of hospital readmission The system uses multiple AI agents coordinated by a supervisor agent.
How we built it Doctorlib uses a multi-agent AI architecture.
FastAPI for backend APIs LangGraph + LangChain for AI workflows OpenAI GPT-4o for reasoning and summarization ChromaDB for medical knowledge retrieval (RAG) SQLite for patient data Streamlit for the frontend dashboard Challenges we ran into Healthcare AI requires high reliability and safety. We ensured the system only provides assistive recommendations and keeps humans in the loop. Another challenge was reducing LLM hallucinations, which we addressed using RAG with clinical guidelines.
Accomplishments that we're proud of Built a multi-agent healthcare AI system Implemented RAG-based medical knowledge retrieval Created clinician dashboards and monitoring workflows Added human-in-the-loop safety checks What we learned We learned how to design multi-agent AI workflows, integrate vector databases with LLMs, and build systems that assist professionals in high-risk domains like healthcare.
What's next for Doctorlib Next steps include EHR integration (FHIR), wearable health monitoring, improved readmission risk models, and deploying the platform on secure cloud infrastructure.
Built With
- llm
Log in or sign up for Devpost to join the conversation.