A multi-paradigm EEG dataset for studying upper limb rehabilitation exercises

Abstract

Most stroke survivors experience persistent upper limb motor dysfunction, and brain-computer interface (BCI) rehabilitation technologies have been widely explored to address this issue. However, systematic comparisons and analyses of differences among rehabilitation paradigms remain challenging due to the lack of multi-paradigm EEG datasets from the same subjects. This study aims to construct an EEG dataset that collects various rehabilitation paradigms for the same subjects. A total of 28 healthy subjects were recruited, and EEG data were collected under six types of upper limb rehabilitation paradigms. Each paradigm involves two or three actions, including grasping and releasing with the left, right, or both hands. The dataset includes both raw EEG signals and preprocessed versions with bandpass filtering and artifact removal. This resource will support studies comparing the neural mechanisms underlying different rehabilitation paradigms and aid in the development of optimized rehabilitation strategies.