JU-PDDv1 EEG Dataset for Parkinson's Disease

The dataset integrates six publicly available EEG datasets collected using different acquisition protocols, electrode configurations, and recording environments into a unified benchmark. To minimize inter-dataset heterogeneity, all recordings underwent a standardized preprocessing pipeline.

If you use JU-PDDv1 in your research, please cite the following: 10.1016/j.bnd.2026.07.003 Each .npy file corresponds to one subject. Each file has the shape (number of epochs, 29, 500) where - Number of epochs: 55 (2-second epochs extracted from a 110-second recording) - 29: common harmonized EEG channels - 500: time samples per epoch (2 seconds × 250 Hz) JU-PDDv1 is intended to support the development and fair evaluation of EEG-based machine learning and deep learning models under standardized subject-independent and cross-dataset evaluation protocols.