A Dataset for Engagement Prediction in Neuromotor Disorder Patients during Rehabilitation

This dataset is related to the work titled "Artificial Intelligence Tools for Engagement Prediction in Neuromotor Disorder Patients undergoing Robot-Assisted Rehabilitation,"  whose aim is to methodologically explore the performance of artificial intelligence algorithms applied to structured datasets made of heart rate variability (HRV) and electrodermal activity (EDA) features to predict the level of patient engagement during robot-assisted gait rehabilitation (RAGR).

It is composed of three Excel files related to the 3-minute windows data augmentation scenario applied to the bimodal dataset made of 14 HRV and 19 EDA features. Specifically:

• ds_bimodal_win_3min.xlsx contains the features extracted from 3-minute windows of HRV and EDA signals, recorded during the RAGR activity, and normalized with respect to the reference (baseline) signals. Features are not z-scored.
• labels_self_win_3min contains one single column with, for each row, the label related to the self-perceived engagement classification target. 
• labels_therapist_win_3min contains one single column with, for each row, the label related to the therapist-perceived engagement classification target.

The coding of classes for both classification targets is:

• 0: Underchallenged
• 1: Minimally Challenged
• 2: Challenged