πŸ₯ TransplantGuard AI

Multi-Agent Medical Guidance System

Live on Google Cloud Run

AI-Powered Support for Transplant Patients

Five specialized AI agents working together to provide instant, expert-level medical guidanceβ€” powered by Google ADK, Gemini 2.0 Flash, and deployed on Cloud Run.

Our AI Medical Team

🎯

Coordinator

Intelligent routing to appropriate specialists

πŸ’Š

Medication Advisor

Missed dose timing and adherence guidance

🩺

Symptom Monitor

Rejection risk assessment from symptoms

⚠️

Drug Interaction

Safety checking for medications and foods

πŸ“Š

Rejection Risk

Evidence-based analysis with SRTR data

Technology Stack

Google Cloud Run Google ADK Gemini 2.0 Flash Firestore Python 3.12 Flask

API Endpoints

GET
/health

Check system health and agent status

POST
/medications/missed-dose

Analyze missed medication doses

POST
/rejection/analyze

Assess rejection risk from symptoms

5
AI Agents
2-3s
Response Time
156
Tests Passing
94.8%
Code Coverage
3
Architectures Evaluated

Evaluated Pub/Sub, In-Process, and ADK patterns before selecting Google ADK sub-agents. View comparison β†’

Try It Out

🌐 Try Interactive Web Demo

Or use the curl commands below to test the API directly

1. Missed Dose Analysis

Test the medication advisor agent - analyzes timing and provides guidance:

curl -X POST https://missed-dose-service-64rz4skmdq-uc.a.run.app/medications/missed-dose \ -H "Content-Type: application/json" \ -d '{"medication":"tacrolimus","scheduled_time":"8:00 AM","current_time":"2:00 PM","patient_id":"demo_patient"}'

Response includes: AI reasoning β€’ Medical recommendation β€’ Risk assessment β€’ SRTR data β€’ Next steps

2. Rejection Risk Analysis

Test the rejection risk agent - evaluates symptoms using real SRTR transplant outcomes data:

curl -X POST https://missed-dose-service-64rz4skmdq-uc.a.run.app/rejection/analyze \ -H "Content-Type: application/json" \ -d '{"symptoms":{"fever":true,"decreased_urine_output":true,"tenderness":true,"fever_temperature":101.5},"patient_id":"demo_patient"}'

Response includes: Urgency level β€’ Rejection probability β€’ Population statistics β€’ Recommended actions