This Graph RAG Hospital chatbot is designed to provide quick answers to various questions about a large hospital system, following questions are listed below:
- What is the current wait time at XYZ hospital?
- Which hospital has the shortest wait time right now?
- At which hospitals are patients reporting issues with billing and insurance?
- Have there been any complaints about cleanliness at the hospitals?
- What feedback have patients given about the communication of doctors and nurses?
- What are patients saying about the nursing staff at XYZ hospital?
- How many visits are currently open and what is their average duration in days?
- Which physician has the shortest average visit duration in days?
- How much was billed for patient 789’s stay?
- Which hospital treated the most Cigna patients in 2023?
- What is the average billing amount for emergency visits by hospital?
This chatbot utilizes Large Language Models (LLM), Retrieval-Augmented Generation (RAG) techniques, and Graph Database (Neo4j), drawing on various data sources to answer these questions.
This project features a versatile chatbot capable of intelligently selecting the appropriate tools to deliver accurate and relevant responses. The chatbot uses a combination of techniques to interact with external APIs, query graph databases, and perform semantic searches within a vector database.
- Agent-Based Tool Selection: The chatbot employs an LLM agent to decide the best tool to use for each query.
- Function Calling with External APIs: It can call external APIs to fetch the necessary data.
- Cypher Query Generation: It generates Cypher query languages to extract information from a graph database using LLMs.
- RAG Technology for Semantic Search: It uses Retrieval-Augmented Generation (RAG) technology to perform semantic searches within the vector database.
- LLM for Response Generation: It utilizes OpenAI LLMs to generate appropriate and coherent responses based on the retrieved information.
- Agent-Based Decision Making: The chatbot's agent evaluates the user's query and determines the most suitable tool to use.
- Retrieval: This chatbot can retrieve relavant information from both structured and unstructured text sources.
2.1. Structured: When a semantic search is needed, the agent uses OpenAI embedding models to search within the vector database and retrieve the most relevant information.
2.2. Unstructured: For queries that can be answered using the graph database, the agent will generate and execute Cypher queries. - External API Calls: If the query requires data from an external source, the agent will call the relevant API and fetch the data.
- RAG: When semantic search or API calling is finished, the agent uses RAG technology to feed the relevant retrieved information and user question into LLM.
- Response Generation: The chatbot then uses the LLM to generate a well-formed response based on retrieved information and user question.
To run this project, follow these steps:
-
Clone the repository:
git clone https://github.com/javaidiqbal11/Graph-RAG-Hospital-Chatbot.git
-
Navigate to the project directory:
cd Graph-RAG-Hospital-Chatbot -
Set up a virtual environment:
python -m venv env source env/bin/activate -
Install dependencies:
pip install -r requirements.txt