Healthcare Information Chatbot

Lilly Moreau & James Power

Project Overview

This project aims to provide users with easy access to detailed information about Louisiana Medicaid, U.S. government Medicare, and general health insurance questions through a chatbot interface. The chatbot is powered by OpenAI's GPT-3.5 model and utilizes LangChain for efficient data retrieval and response generation, with a Chroma database storing pre-processed, vector-embedded document data. Our Django backend handles API requests and data management, while the React frontend offers an intuitive user interface styled with Material UI.

Key Features

• Information Retrieval: Dynamically fetches relevant information from a curated database.
• Chat Interface: Users can query the system in natural language.
• Responsive Design: The UI adjusts for different devices, enhancing accessibility.
• Data Sources: Compiles reputable sources to ensure accuracy and reliability.

Technical Stack

• Backend: Django
• Frontend: React with Material UI
• AI and NLP: OpenAI GPT-3.5, LangChain LLM
• Database: Chroma (Vector database for efficient data retrieval)

Key Components

Backend

• chat_chain.py: Central to the chatbot's operation, orchestrating the data retrieval, processing, and interaction with GPT-3.5.
• pull_data.py: Responsible for fetching and preprocessing data from reputable sources to populate the Chroma DB.
• chatbot_post.py: Hanldes the post request and invokes the chat chain
• Environment Variables: Utilizes .env for managing sensitive API keys securely. #### Frontend
• chat.jsx: Implements the chat interface, handling user inputs and displaying chatbot responses.
• PersistentDrawer.jsx: Manages the navigation drawer, theme toggling between light and dark modes, and the switching of data sources.

Built With

• chroma
• django
• javascript
• langchain
• materials-ui
• openai
• python
• react