This dataset was recorded during sessions with externalized Deep Brain Stimulation (DBS) Essential Tremor (ET) patients. Patients were asked to perform a pegboard insertion task, a posture hold with arms stretched out front, and a rice pouring task between two cups. Data was recorded both with DBS stimulation on and off. Here we only provide a brief description and refer to the Methods section of [1] for further details on the patients (incl. clinical information), hardware, and recording setup.

The dataset is structured as a series of 8 folders (1, 2, …, 8), with each folder corresponding to data recorded from an individual patient. Each of these folders contains HDF5 files named <task>_<stimulation_condition>.h5 (e.g. Pegboard_on.h5), where each file consist of arrays containing bilateral, multi-channel LFPs under the “LFP” key and a corresponding movement label under the “label” key. Movement labels were extracted from EMG and accelerometer data attached to the arms and hands of patients. The LFP data comes highpass-filtered (4th order, one-pass butterworth filter with cut-off frequency of 1 Hz). The sampling rate for all datasets is 2048 Hz.

This dataset has been set up to directly work with the code provided at https://github.com/fer-rplazas/feature-extraction-methods, which includes utilities to load the data, train models with it, and generate figures.