Parkinson's EEG Analysis & Counterfactual Explanations

This project implements a machine learning pipeline for detecting Parkinson's Disease (PD) from EEG signals. It utilizes Multivariate Direct Mode Decomposition (MDMD) for feature extraction and a Linear SVM for classification. Furthermore, it employs Counterfactual Explanations (using the Native Guide method) to provide interpretable insights into why a specific segment was classified as PD.

Pipeline Overview

1. Data Loading & Preprocessing:
   • Loads EEG data in BIDS format using mne.
   • Applies notch filtering (50/60Hz) and resamples to 512Hz.
   • Segments data into 2-second non-overlapping windows.
2. Signal Decomposition (MDMD):
   • Constructs multichannel Hankel trajectory matrices.
   • Performs SVD to decompose signals into intrinsic Modes (Trends/Oscillations).
   • Reconstructs signals using Diagonal Averaging.
3. Feature Extraction:
   • Extracts 3 features per mode/channel: Frequency ($f_q$), Power ($P_q$), Amplitude ($S_q$).
4. Classification:
   • Trains a Linear Support Vector Machine (SVM) with Stratified K-Fold Cross-Validation.
   • Optimizes hyperparameters using Grid Search.
5. Explainability (XAI):
   • Generates Counterfactuals: Finds the minimal changes required to flip a "Parkinson's" prediction to "Healthy."
   • Uses the Native Guide method: Interpolates the patient's data towards the nearest existing "Healthy" sample to ensure biological plausibility.

Prerequisites

Ensure you have Python 3.8+ and the following libraries installed:

pip install numpy pandas scipy scikit-learn matplotlib mne joblib