JU-PDDv1 EEG Dataset for Parkinson's Disease
Description
JU-PDDv1 is a harmonized multi-source EEG dataset developed to facilitate reproducible and generalizable Parkinson's disease (PD) research. 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.
Files
Steps to reproduce
Please refer to this paper: 10.1016/j.bnd.2026.07.003
Institutions
- Jadavpur UniversityWest Bengal, Kolkata