AI Diabetes Diagnosis System

This repo is the complete code implementation of our project, AI-Powered Conversational Diabetes Diagnosis System.

Repo introduction

The repo includes the original code and the final code completion of this project. Users only need the "local" and "server" folders for the deployment.

How to use the repo?

1. Download or fork the repo to your local device.
2. Install or update the Python package using the requirement_gpu_machine.txt on your GPU machine and requirement_localhost.txt on your local device. The Python version we used is 3.9.

   pip install requirement_gpu_machine.txt

   pip install requirement_localhost.txt

4. Move the "server" folder to your GPU machine, and make sure the machine has 30 Gib GPU RAM. If your local device has the capability, you can keep the "server" folder in your local device.
5. Use Python to run the ai_nurse_llm.py on your GPU machine. This step needs to take some time.

   python "Path to the folder"/server/ai_nurse_llm.py

6. If the GPU machine and the localhost are the same device, it is fine. If not, please let the device with the "local" folder SSH connect to the port (default:5000) in your GPU machine.

   ssh -L 5000:localhost:5000 "your GPU machine account and address"

7. Move the "local" folder to your localhost.
8. On your localhost, open the terminal and move to the address of the "local" folder, then run the following code.

    uvicorn diabetes_server:app --reload

9. Open a new terminal and move to the address of the "local" folder, then run the following code.

    python -m http.server 8080

10. Use the Chrome or Edge browser to open the link: http://localhost:8080/
11. After all steps go well, you can talk with our AI nurse now.