A real-time, multi-agent healthcare assistant built to intelligently analyze patient symptoms and route cases to the appropriate medical specialists. This system combines large language models (LLMs), a Streamlit web interface, and MongoDB for an interactive triage experience with specialist alerts and diagnostic support.
AIATL leverages CrewAI, Streamlit, and MongoDB Atlas to simulate a collaborative team of medical AI agents capable of:
- Interpreting patient-reported symptoms and history
- Classifying and routing cases to relevant specialists (cardiology, neurology, pulmonology)
- Notifying appropriate doctors through in-app alerts
- Reducing demographic bias in care through RAG-enhanced diagnosis agents
- CrewAI — Multi-agent LLM framework for diagnosis and routing
- Streamlit — Frontend for patient and doctor interface
- MongoDB Atlas — Cloud database for storing patient history and doctor feedback
- GCP — Infrastructure and deployment
- Python — Core development language
-
🧑⚕️ 6-Agent CrewAI System
Each agent is responsible for stages like symptom analysis, specialty assignment, and diagnosis generation. -
⚡ <10s Specialist Routing
Designed to classify and forward patient cases to the correct specialist with minimal delay. -
📊 Real-Time Interface
Patients enter symptoms via a user-friendly Streamlit frontend; doctors are notified with alerts inside their dashboard. -
🧠 RAG-Enabled Diagnosis
Retrieval-Augmented Generation agents incorporate medical knowledge to improve diagnostic accuracy and reduce bias. -
✅ 95% Routing Accuracy
Based on internal tests, the triage engine achieves high classification accuracy across specialties.