Drai Medical Chatbot

Welcome to the Drai Medical Chatbot! This application leverages the OpenAI GPT-3.5 Turbo model that has been fine-tuned to simulate a medical assistant capable of providing preliminary medical advice. Users can interact with the chatbot by sending messages describing their symptoms or asking health-related questions.

Getting Started

To get started with the chatbot, follow these steps:

1. Clone the repository:

   git clone https://github.com/MBobie/Drai.git
   cd drai

2. Install dependencies:

   pip install -r requirements.txt

3. Set up your environment variables:

   Create a .env file in the project root and add your OpenAI API key:

   OPENAI_API_KEY=your_openai_api_key
   

4. Run the FastAPI application:

   uvicorn main:app --reload

   The application will be accessible at http://127.0.0.1:8000.

Usage

Swager UI

Swagger UI provides a convenient way to test your API endpoints. Visit http://127.0.0.1:8000/docs in your browser to explore and interact with the API.

Chat Endpoint

Send a POST request to /chat/ with a JSON payload containing the user's message. The Swagger UI allows you to input and test different requests.

OR

Send a POST request to /chat/ with a JSON payload containing the user's message:

curl -X POST "http://127.0.0.1:8000/chat/" -H "accept: application/json" -H "Content-Type: application/json" -d '{"message": "I have a headache"}'

Special Commands

The chatbot recognizes special commands such as exit, clear, help, and info. Use these commands to interact with the chatbot:

/exit: End the conversation.

/clear: Clear the conversation history and start a new session.

/help: Get information on using the chatbot.

/info: Learn more about the chatbot and its capabilities.

Conversation History

To retrieve the conversation history, send a GET request to /history/ using the Swagger UI

OR

curl -X GET "http://127.0.0.1:8000/history/" -H "accept: application/json"

Test

To run the tests in the script

pytest test_main.py

Feedback

Your feedback is valuable! Feel free to provide input on the chatbot responses to help improve its performance.

Acknowledgments

This project utilizes the OpenAI GPT-3.5 Turbo model for natural language processing.

Happy chatting with DRAI!

Credits

David Aleoghena Victor Ofor