EEG Alzheimer's Detection Using LLM and ML Classifier

This project is designed to analyze EEG data and predict the likelihood of Alzheimer's disease using a combination of a Large Language Model (LLM) and a Machine Learning (ML) classifier. The tool provides an interface for uploading EEG data and receiving diagnostic predictions.

Features

EEG Data Upload: Easily upload EEG files for analysis.

Integrated Analysis: Utilizes both LLM and ML models to enhance prediction accuracy.

User Interface: Built with the "gradio" package for a seamless user experience.

Integration with OpenAI and NVIDIA: This project utilizes NVIDIA's Palmyra-Med-70B-32K API in collaboration with OpenAI's advanced language models to enhance the analysis and prediction capabilities of the system.

Installation

Clone the Repository:

git clone https://github.com/yourusername/your-repo-name.git
cd your-repo-name

Install Dependencies:

Ensure you have Python installed. Install the necessary packages using the provided text file:

pip install -r requirements.txt

Set Up API Key:

Obtain an API key from nVIdia by making an account. Create a .env file in the project folder and add your API key:

SECRET_API_KEY=your_api_key_here

Usage Download EEG Files: Obtain the EEG files from the "Data" folder in this repository. s Modify the code: Change the ML Path to your the path on your local PC in the main code script.

Run the Main Function: Execute the main script to start the application.

Access the User Interface: Upon running the main function, a link will be generated. Follow this link to access the user interface.

Navigate to the Analysis Page: The link will direct you to the analysis page.

Input EEG File Path: In the "File Path" section at the top, enter the local path to your EEG file. The path should resemble the following format: downloads\Data\sub-001_task-eyesclosed_eeg.set

Analyze the Data: Click the "Analyze EEG Data" button, which is highlighted in orange.

View the Results: The output section will display the LLM prompt with the analysis results.

Contributing

We welcome contributions from the community. Please feel free to submit issues or pull requests.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Contact

For questions or support, please contact Soudeh (Venus) Mostaghimi at [email protected], Trisha Mendoza at [email protected], or Blanca Romer at [email protected].