EMG Based Body-Machine Interface for Adaptive and Personalized Robotic Training of Persons with Multiple Sclerosis

Abstract

Multiple sclerosis is a complex neurological disease that results in motor impairment associated with muscle weakness and lack of motor coordination. Indeed, previous studies showed that, while activities in isolated arm muscles appeared generally similar to those of unimpaired subjects, shoulder muscle coordination with arm motions was affected by MS and there was a marked co-activation of the biceps and triceps in the extension movements. This inability to activate muscles independently has a significant impact in motor function therefore reducing the co-contraction could improve the overall arm function. In this pilot study, we developed a body-machine interface based on muscle activities with the goal of ‘breaking’ the abnormal triceps-biceps co-activation during planar flexion-extension movements of people with multiple sclerosis during a robot-based task. The task consisted in 2D center-out reaching movements with the assistance of a robotic manipulandum. When the subject was not exhibiting the abnormal triceps-biceps co-activation for three consecutive movements the robot was decreasing the assistance. Subjects trained for up to six 1-hour sessions in three weeks. Results showed that the assistance from the robot decreased within each session for most of the subjects, while the movement became faster and straighter. The comparison between muscle activity before and after the training with this body-machine interface demonstrated that subjects learned how to reduce the triceps-biceps co-activation.