What problem does it solve? Processing the SEED-VIG vigilance EEG dataset requires coordinating BIDS validation, EEG preprocessing, spectral feature extraction, and classification across multiple tools, which is error-prone when done manually. ## Core Features & Use Cases - BIDS Validation: Checks SEED-VIG directory structure, subject completeness, EEG file presence, and vigilance label availability, producing a CSV compliance report. - Feature Extraction: Computes band power (delta through gamma) and vigilance spectral ratios (theta/alpha, (theta+alpha)/beta, theta/beta) per epoch from .mat, .edf, .set, and .vhdr EEG files. - Vigilance Classification: Trains SVM and Random Forest classifiers with Leave-One-Subject-Out cross-validation for alert vs. drowsy detection. - Use Case: A researcher downloads the 23-subject SEED-VIG dataset and wants an end-to-end pipeline that validates the BIDS layout, extracts spectral features, and reports classification accuracy and AUC. ## Quick Start Run the full SEED-VIG pipeline on my local BIDS dataset to validate the data, extract EEG features, and classify vigilance states.
