Estimation of features informativeness of the EMG signal in the problem of forearm prosthesis controlling

Abstract

In this study estimation of features informativeness of the EMG signal in the problem of forearm prosthesis controlling is considered. Logistic regression is used to classify six hand movements: grasping, opening, flexion, pronation and supination using fourteen time domain features. To evaluate the importance of features linear regression coefficients are used. Most informative features are shown.