A demo and proof-of-concept for using Large Language Models to generate personalized care plans with human-in-the-loop clinical oversight
See demo live at here
This proof-of-concept showcases how Large Language Models can assist healthcare providers in generating personalized care plans while maintaining essential clinical oversight. The system demonstrates a human-in-the-loop approach where AI generates initial care plans that require clinician review and approval before implementation.
- AI-Powered Generation: Uses OpenAI GPT models to create initial care plans from patient data
- Clinical Review Workflow: All AI-generated plans require clinician approval before activation
- Batch Processing: Demonstrates scalable processing of multiple patient records
- Role-Based Access: Separate interfaces for patients, clinicians, and administrators
- Patient Data Input: Upload patient information via CSV or manual entry
- AI Generation: LLM processes patient data to create initial care plans
- Clinical Review: Clinicians review, modify, and approve AI-generated plans
- Patient Access: Approved plans become available in patient portal
Frontend: React + TypeScript + Material-UI
Backend: FastAPI + Python + Pydantic
AI/ML: OpenAI GPT + FAISS Vector Search
Infrastructure: Docker + AWS ECS
See ARCHITECTURE.md for detailed technical specifications and system design.