Skip to content
/ RavenCare Public

Built for Composio x DevClub IITD x Chirality Labs Agents in Production Hackathon

License

GPL-3.0 license
3 stars 1 fork Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

SamirXR/RavenCare

BranchesTags

Repository files navigation

RavenCare : AI-Powered Medical Triage System Built with Composio

Overview

RavenCare is a production-ready, AI-powered medical triage system that revolutionizes patient assessment through hybrid AI consensus methodology, intelligent doctor matching, and seamless healthcare workflow automation. Powered by Composio for enterprise-grade Google Workspace integrations.

What Makes RavenCare Special?

  • Ensemble Intelligence Architecture: Three AI models (Gemini, Grok, OpenAI) collaborate through consensus-based decision making for enhanced diagnostic accuracy
  • Intelligent Matching Algorithm: 8-factor scoring system with 200+ medical keywords for precise doctor-patient alignment
  • Complete Automation: End-to-end workflow automation from triage to appointment scheduling via Composio
  • Production-Grade: Finetune two 7b LLMs such as Mistral or Gemma and deepseek's reasoning Model! this will keep the data secure

Read : information.md to see how it works.

Key Features

Ensemble Intelligence Architecture

  • Gemini 2.5 Pro: Comprehensive symptom analysis and medical specialty mapping
  • Grok 4 Fast Reasoning: Urgency scoring (0-100) and risk assessment with red flags
  • OpenAI O4-Mini: Final clinical evaluation and cross-validation

Advanced Doctor Matching (8-Factor Algorithm)

  • Sub-Specialization Matching: 200+ keyword database for precise expertise alignment
  • Urgency-Based Prioritization: Critical cases mapped to most experienced doctors
  • Age-Appropriate Care: Pediatric and geriatric specialization awareness
  • Smart Availability: Slot matching with urgency multipliers
  • Language Preferences: Multi-language support
  • Quality Ratings: Doctor ratings and awards consideration
  • Match Transparency: Full scoring breakdown and explanations

Supported Medical Specialties

Cardiology • Gastroenterology • Hepatology • Neurology • Orthopedics • Pediatrics • Dermatology • Ophthalmology • ENT • Psychiatry • Pulmonology

Each specialty includes sub-specialization detection for precise matching.

Quick Start

Prerequisites

- Python 3.8+
- Internet connection (for AI APIs)
- Google account (for Composio integrations)
- Composio account (free tier available)
- Git (for cloning repository)

Installation

1. Clone the Repository

# Clone from GitHub
git clone https://github.com/SamirXR/RavenCare.git

# Navigate to project directory
cd RavenCare

2. Install Dependencies

pip install -r requirements.txt

3. Configure Environment Variables

# Copy template
copy .env.example .env

# Edit with your API keys
notepad .env

4. Verify Configuration

python -c "from src.config import config; config.print_config_status()"

Running the System

Web Dashboard (Recommended)

python app.py

Open http://localhost:5000 in your browser

Command Line

python -m src.triage_orchestrator

Configuration

Required API Keys

RavenCare requires several API keys for full functionality. Below is a detailed guide on obtaining each key:

1. Gemini API Key (Google AI Studio)

GEMINI_API_KEY=your_gemini_key

How to obtain:

  1. Visit Google AI Studio
  2. Sign in with your Google account
  3. Click "Get API Key" or "Create API Key"
  4. Copy the generated key and paste it in your .env file

Free Tier: Available with generous quota for testing

2. Grok API (Azure AI Services)

GROK_API_KEY=your_grok_key
GROK_ENDPOINT=https://your-endpoint.services.ai.azure.com/openai/v1/

How to obtain:

  1. Create an account at Azure Portal
  2. Navigate to Azure AI Services or Azure OpenAI
  3. Click "Create" to set up a new resource
  4. Once created, go to "Keys and Endpoint" in the left sidebar
  5. Copy KEY 1 or KEY 2 → Use as GROK_API_KEY
  6. Copy the Endpoint URL → Use as GROK_ENDPOINT
  7. Go to "Model deployments" and deploy the Grok model
  8. Note the deployment name for GROK_MODEL_NAME

Note: Requires Azure subscription (free trial available)

3. OpenAI API (Azure OpenAI Service)

OPENAI_API_KEY=your_openai_key
OPENAI_ENDPOINT=https://your-endpoint.cognitiveservices.azure.com/

How to obtain:

  1. Go to Azure Portal
  2. Search for "Azure OpenAI" service
  3. Create a new Azure OpenAI resource (may require access approval)
  4. Once approved and created, navigate to "Keys and Endpoint"
  5. Copy KEY 1 or KEY 2 → Use as OPENAI_API_KEY
  6. Copy the Endpoint URL → Use as OPENAI_ENDPOINT
  7. Deploy the o4-mini model in "Model deployments"

Important: Azure OpenAI access may require application approval. Apply at Azure OpenAI Access Form

4. Composio Platform Keys

COMPOSIO_API_KEY=your_composio_key
COMPOSIO_USER_ID=your_user_id

How to obtain:

  1. Sign up at Composio Dashboard
  2. Navigate to Settings → API Keys
  3. Copy your API Key → Use as COMPOSIO_API_KEY
  4. Find your User ID in the dashboard or profile → Use as COMPOSIO_USER_ID

Free Tier: Available with sufficient quota for development

5. Composio Connected Accounts (Optional - Enables Automation)

COMPOSIO_SHEETS_ACCOUNT_ID=your_sheets_account
COMPOSIO_CALENDAR_ACCOUNT_ID=your_calendar_account
COMPOSIO_GMAIL_ACCOUNT_ID=your_gmail_account
COMPOSIO_DRIVE_ACCOUNT_ID=your_drive_account

How to obtain:

  1. Go to Composio Apps
  2. Connect each Google Workspace app:
    • Google Sheets (for data dashboards)
    • Google Calendar (for appointment scheduling)
    • Gmail (for email notifications)
    • Google Drive (for PDF report storage)
  3. Authorize each app with your Google account
  4. After connection, go to "Connected Accounts"
  5. Copy each Account ID and add to .env

Note: These are optional but highly recommended for full automation features

6. Application Settings

ADMIN_EMAIL=admin@your-domain.com
FLASK_SECRET_KEY=random_secret_key
  • ADMIN_EMAIL: Your email to receive system notifications
  • FLASK_SECRET_KEY: Generate securely with: 
    python -c "import secrets; print(secrets.token_hex(32))"

Quick Setup Checklist

  • Create .env file from .env.example
  • Add Gemini API key (Google AI Studio)
  • Add Azure Grok credentials (Key + Endpoint)
  • Add Azure OpenAI credentials (Key + Endpoint)
  • Add Composio API key and User ID
  • (Optional) Connect Google Workspace apps in Composio
  • Set admin email
  • Generate Flask secret key
  • Test configuration: python -c "from src.config import config; config.print_config_status()"

Composio Integrations

Composio powers all Google Workspace integrations for seamless automation.

Gmail Integration

File: src/services/email_service.py

Features:

  • HTML-formatted professional emails
  • Multi-recipient support (admin, patients, doctors)
  • Embedded calendar links
  • Delivery tracking

Composio Actions Used:

  • GMAIL_SEND_EMAIL: Send notifications

Google Sheets Integration

File: src/services/sheets_service.py

Features:

  • Real-time online dashboards
  • Formatted triage data tables
  • Auto-generated reports
  • Shareable links

Composio Actions Used:

  • GOOGLESHEETS_SHEET_FROM_JSON: Create sheets from data

Google Calendar Integration

File: src/services/calendar_service.py

Features:

  • Automatic appointment scheduling
  • Meeting link generation
  • Multi-participant events
  • Timezone handling

Composio Actions Used:

  • GOOGLECALENDAR_CREATE_EVENT: Schedule appointments

Google Drive Integration

File: src/services/email_service.py

Features:

  • PDF report storage
  • Shareable link generation
  • Public access permissions
  • Organized folder structure

Composio Actions Used:

  • GOOGLEDRIVE_UPLOAD_FILE: Upload PDFs
  • GOOGLEDRIVE_ADD_FILE_SHARING_PREFERENCE: Make files shareable

Web Dashboard

Real-time Streaming Interface

Features:

  • Live progress updates via Server-Sent Events (SSE)
  • Visual progress bars

Access: Run python app.py then navigate to http://localhost:5000

Use Cases

  • Hospital Emergency Departments: Rapid patient prioritization
  • Telemedicine Platforms: Remote triage and consultation
  • Healthcare Chatbots: Automated initial assessment
  • Medical Hotlines: Operator decision support
  • Walk-in Clinics: Efficient patient routing
  • Healthcare Education: Teaching triage principles

Disclaimer

This system is designed for educational and demonstration purposes. It should not replace professional medical judgment. Always consult qualified healthcare professionals for medical advice.

Future Roadmap

  • Custom Finetuned Model learning for outcome prediction
  • Multi-language patient support
  • EHR system integration
  • Mobile application
  • Advanced analytics dashboard
  • Prescription management

About

Built for Composio x DevClub IITD x Chirality Labs Agents in Production Hackathon

Resources

Readme

License

GPL-3.0 license
Activity

Stars

3 stars

Watchers

0 watching

Forks

1 fork
Report repository

Releases

No releases published

Packages

No packages published

Languages