🏥 AI Healthcare Bot with Memory

An intelligent healthcare assistant with memory capabilities using Mem0, built using FastAPI, Chainlit, and mem0. This bot can remember patient information across conversations, providing personalized healthcare support.

Demo

mem0_demo.mp4

🌟 Features

• 🧠 Memory System: Remembers patient information and conversation history using mem0
• 🔄 Streaming Responses: Real-time streaming of AI responses for a better user experience
• 🏥 Healthcare Focus: Specialized in healthcare-related conversations and information
• 🌐 Dual Interface: API backend with FastAPI and interactive UI with Chainlit
• 🔒 Secure: Designed with healthcare data privacy considerations
• 📱 Responsive UI: Modern, user-friendly interface for patient interactions

🚀 Quick Start

1. Installation

git clone https://github.com/extrawest/ai_healthcare_bot_with_memory.git
cd ai_healthcare_bot_with_memory

python -m venv .venv
source .venv/bin/activate

pip install -r requirements.txt

2. Configuration

Copy the example environment file and configure your settings:

cp .env.example .env

Edit the .env file with your own settings:

• Set your OpenAI API key
• Configure Qdrant settings for vector storage
• Adjust host/port settings if needed

3. Start the FastAPI Backend

First, start the FastAPI backend server:

python -m src.main

The API will be available at http://0.0.0.0:8000 by default.

4. Start the Chainlit UI

In a new terminal window (with the virtual environment activated), start the Chainlit UI:

python run_chainlit.py

The UI will be available at http://0.0.0.0:8001 by default.

5. Interact with the Bot

Open your browser and navigate to http://0.0.0.0:8001 to start interacting with the healthcare bot. The bot will remember your conversation history and provide personalized healthcare support.

💻 Usage

API Endpoints

Chat Completions

POST /chat/completions

Request body:

{
  "user_message": "I have a headache and would like to schedule an appointment",
  "user_id": "user_12345"
}

Response: A streaming response with AI-generated content.

Memory System

The bot uses mem0 to store and retrieve relevant information from previous conversations. This allows it to:

• Remember patient symptoms and conditions
• Recall previous appointments and discussions
• Provide personalized healthcare advice based on patient history

📋 Requirements

• Python 3.9+
• OpenAI API key
• Qdrant instance (for vector storage)
• FastAPI
• Chainlit
• mem0
• langchain

🔧 Technical Implementation

Memory Configuration

The system uses mem0 for memory storage with the following configuration:

• OpenAI embeddings (text-embedding-3-large)
• Qdrant vector store for efficient retrieval
• Custom prompt for healthcare entity extraction

Streaming Implementation

The application implements streaming responses using:

• FastAPI's StreamingResponse
• Custom streaming service that converts regular responses to streams
• Chainlit's streaming capabilities for real-time UI updates

Healthcare Support Agent

The AIHealthcareSupport agent:

• Retrieves relevant memories for each query
• Builds context from previous conversations
• Uses OpenAI's GPT-4o model for generating responses
• Stores new conversation data in memory

Developed by extrawest. Software development company