Proactive Follow-Up for High-Risk Patients

📌 Description

This project aims to reduce rehospitalization risks for high-risk patients (post-operative, chronic diseases, postpartum, etc.) through a proactive artificial intelligence system. Doctors can configure a custom AI agent based on the patient's pathology, particularly in geriatrics, where multiple agents collaborate for optimal monitoring.

🏆 Winner of the Doctolib AI x One Health Hackathon 2025

(Djason Gadiou / Adrien Ezerzer / Louis Martinez / Jacques Sun)

Overview

This AI-powered system enhances interactions between doctors and patients by integrating:

• Web research for relevant medical insights
• Retrieval-Augmented Generation (RAG) for intelligent responses
• Text-to-speech capabilities for communication
• Automated patient record management and reporting

Workflow Diagram

graph TD
    subgraph Doctor Interaction
        A[Doctor] -->|Configures Custom AI Agent| B[ElizaOS on Scaleway]
    end

    subgraph AI Agent Operations
        B -->|Generates Specific AI Agent| C[AI Agent via ElizaOS]
        C -->|Performs Web Research| D[Web Resources]
        C -->|Utilizes RAG Model| E[Retrieval-Augmented Generation]
        E -->|Stored/Retrieved from| G[Supabase Database]
        C -->|Text-to-Speech| F[ElevenLabs]
        C -->|Updates Records & Creates Report| M[Patient Records & Report]
        M -->|Stored in| G
    end

    subgraph Patient Interaction
        C -->|Initiates Call| H[Patient]
        H -->|Provides Responses| C
    end

    subgraph Alert System
        C -->|Sends Alert| I[Doctor]
        C -->|Suggests| J[Patient Calls Emergency Services]
        I -->|Confirms Alert| K[Schedule Appointment]
        I -->|Confirms Alert| L[Call Patient]
    end

Loading

Features

Doctor Interaction

• Configure a custom AI agent via ElizaOS hosted on Scaleway.

AI Agent Operations

• Generates AI agents for personalized patient monitoring.
• Conducts web research for relevant medical insights.
• Uses a RAG model for enhanced information retrieval.
• Stores and retrieves patient data from Supabase Database.
• Provides text-to-speech functionality for seamless interaction.
• Updates patient records and generates automated reports.

Patient Interaction

• AI agent initiates calls to check on the patient.
• Patients respond to health inquiries, and AI processes the responses.

Alert System

• AI detects warning signs and sends alerts to doctors.
• Can suggest emergency services if needed.
• Doctors can confirm alerts and take appropriate action:
  • Schedule an appointment
  • Call the patient for further assessment

Data Management

• Supabase Database stores patient records and reports.
• Dynamic retrieval and updates ensure real-time data availability.

How It Works

1. Doctor configures an AI agent for tailored patient follow-up.
2. AI conducts regular health assessments using personalized questions.
3. Intelligent response analysis powered by RAG and web research.
4. Three possible scenarios:
   • ✅ All is well → Normal follow-up.
   • ⚠️ Warning signs detected → Preventive intervention.
   • 🚨 Health deterioration → Immediate alert and human intervention.
5. Proactive medical team validates alerts and schedules appointments as needed.

Key Benefits

✅ Reduced rehospitalization rates through real-time monitoring.
✅ Optimized healthcare professional time with AI automation.
✅ Enhanced patient experience through personalized support.
✅ Seamless automation while maintaining human supervision.

Conclusion

This AI-driven system streamlines doctor-patient interactions, automates record-keeping, and enhances patient safety through real-time alerts and recommendations, ultimately improving healthcare outcomes.

Powered by ElizaOS