Skip to content

README.md

Latest commit

 

History

History
257 lines (191 loc) · 7.58 KB

README.md

File metadata and controls

257 lines (191 loc) · 7.58 KB

Health Chatbot

A healthcare-focused RAG (Retrieval-Augmented Generation) chatbot built with TypeScript, Hono, and AI-powered document processing. This chatbot leverages OpenAI and OpenRouter for intelligent responses, Pinecone for vector similarity search, and PostgreSQL for data persistence.

Features

  • 🤖 AI-Powered Responses: Integration with OpenAI and OpenRouter APIs for intelligent health-related queries
  • 📄 Document Processing: Extract and process healthcare documents (.docx files) into vector embeddings
  • 🔍 Vector Search: Pinecone-based semantic search for relevant medical information
  • 💾 Persistent Storage: PostgreSQL database with Drizzle ORM for data management
  • 🔐 Secure: Security headers, environment validation, and API key management
  • 🌐 REST API: Clean REST API built with Hono framework
  • 📱 Webhook Support: WhatsApp integration via Wasenderapi for multi-channel access
  • 🔄 Document Ingestion: CLI tool for importing healthcare documents into the vector database

Tech Stack

  • Runtime: Node.js with TypeScript
  • Framework: Hono - Ultrafast web framework
  • Database: PostgreSQL with Drizzle ORM
  • Vector DB: Pinecone for semantic search
  • AI Models: OpenAI & OpenRouter APIs
  • Document Processing: Mammoth for .docx extraction
  • Validation: Zod for runtime schema validation
  • Messaging: WhatsApp integration via Wasenderapi

Quick Start

Prerequisites

  • Node.js 18+
  • pnpm (or npm/yarn)
  • PostgreSQL database
  • API Keys:
    • OpenAI
    • OpenRouter
    • Pinecone
    • Wasenderapi (for WhatsApp integration)

Installation

# Clone the repository
git clone https://github.com/Walon-Foundation/health-chatbot.git
cd health-chatbot

# Install dependencies
pnpm install

# Configure environment variables
cp .env.example .env.local
# Edit .env.local with your API keys and database URL

Development

# Start development server
pnpm dev

# Server runs on http://localhost:5000
# API available at http://localhost:5000/api/v1/

Database Setup

# Generate database migrations
pnpm db:generate

# Apply migrations to database
pnpm db:push

# Open Drizzle Studio for database management
pnpm db:studio

Document Ingestion

# Ingest health documents into Pinecone vector database
# Place your .docx files in src/data/ directory first
pnpm ingest-doc

Production

# Build the project
pnpm build

# Start the server
pnpm start

API Endpoints

Health Check

GET /api/v1/

Query Endpoint

POST /api/v1/query
Content-Type: application/json

{
  "query": "What are the symptoms of diabetes?",
  "conversationId": "optional-conversation-id"
}

Webhook

POST /api/v1/webhook

Receives WhatsApp messages and processes them through the chatbot.

Project Structure

.
├── src/
│   ├── config/           # Configuration files for external services
│   │   ├── env.ts        # Environment variables validation
│   │   ├── openai.ts     # OpenAI client setup
│   │   ├── openRouter.ts # OpenRouter client setup
│   │   └── pinecone.ts   # Pinecone client setup
│   ├── db/               # Database configuration
│   │   ├── db.ts         # Database connection
│   │   └── schema.ts     # Drizzle ORM schema
│   ├── lib/              # Utility libraries
│   │   ├── doc-reader.ts # Extract text from .docx files
│   │   ├── doc-splitter.ts # Split documents into chunks
│   │   └── ingest-doc.ts # CLI tool for document ingestion
│   ├── routes/           # API routes
│   │   ├── query.ts      # Query handler
│   │   └── webhook.ts    # WhatsApp webhook handler
│   └── index.ts          # Application entry point
├── drizzle.config.ts     # Drizzle ORM configuration
├── tsconfig.json         # TypeScript configuration
└── package.json          # Dependencies and scripts

Environment Variables

Create a .env.local file with the following variables:

# Database
DATABASE_URL=postgresql://user:password@localhost:5432/health_chatbot

# Environment
NODE_ENV=development

# Pinecone Vector Database
PINE_CONE_API_KEY=your_pinecone_api_key
PINE_CONE_INDEX_NAME=your_index_name

# AI Models
OPENAI_API_KEY=your_openai_key
OPENROUTER_API_KEY=your_openrouter_key

# WhatsApp Integration
WASENDERAPI_API_KEY=your_wasenderapi_key

Development Scripts

Command Description
pnpm dev Start development server with hot reload
pnpm build Build TypeScript to JavaScript
pnpm start Run production build
pnpm ingest-doc Process and ingest healthcare documents
pnpm db:generate Generate database migrations
pnpm db:push Apply migrations to database
pnpm db:migrate Run pending migrations
pnpm db:studio Open Drizzle Studio dashboard

Testing the Bot

Development Testing

Once the server is running and configured locally, you can test the Health Chatbot via WhatsApp:

  1. Send a message to: [WHATSAPP_BOT_NUMBER_PLACEHOLDER]
  2. Type your health query, for example:
    • "What are the symptoms of diabetes?"
    • "How can I reduce high blood pressure?"
    • "What are the side effects of common pain relievers?"
  3. Wait for the response - The bot will search through your ingested healthcare documents and provide an AI-powered response

Note: Make sure your bot number is properly configured in your Wasenderapi dashboard and the webhook is receiving messages correctly.

Production Bot

To test the production bot that's currently live, send a WhatsApp message to:

+23233482361

The production instance is running with optimized settings and ingested healthcare documents. You can ask any health-related questions and receive AI-powered responses based on the knowledge base.

## Note: This temporary

Security

This project implements multiple security measures:

  • Environment Validation: All environment variables are validated at startup using Zod
  • Secure Headers: Hono's secureHeaders middleware prevents common web vulnerabilities
  • API Key Management: Sensitive keys are managed via environment variables (never committed)
  • Database Encryption: PostgreSQL connection uses SSL
  • CORS & Input Validation: Request validation with Zod schemas

See SECURITY.md for detailed security guidelines.

Contributing

We welcome contributions! Please see our Contributing Guidelines for details on how to:

  • Report bugs
  • Suggest features
  • Submit pull requests
  • Set up your development environment

License

This project is licensed under the MIT License - see the LICENSE file for details.

Support

For questions or issues:

  1. Check the SETUP.md guide for setup help
  2. Review SECURITY.md for security concerns
  3. Open an issue on GitHub
  4. Contact the Walon Foundation

Acknowledgments

  • Built by the Walon Foundation
  • Powered by OpenAI, OpenRouter, and Pinecone
  • Built with Hono framework
  • Database powered by PostgreSQL and Drizzle ORM

Roadmap

  • Support for PDF documents
  • Multi-language support
  • User authentication and conversation history
  • Advanced RAG parameters tuning
  • Admin dashboard for document management
  • Rate limiting and usage analytics

Note: This chatbot is for informational purposes. Always consult with healthcare professionals for medical advice.