Upper-Extremity Stroke Therapy Task Discrimination Using Motion Sensors and Electromyography

Abstract

Brain injury resulting from stroke often causes upper-extremity motor deficits that limit activities of daily living. Several therapies being developed for motor rehabilitation after stroke focus on increasing time spent using the extremity to promote motor relearning. Providing a novel system for user-worn therapy may increase the amount and rate of functional motor recovery. A user-worn system comprising accelerometers, gyroscopes, and electromyography amplifiers was used to wirelessly transmit motion and muscle activity from normal and stroke subjects to a computer as they completed five upper-extremity rehabilitation tasks. An algorithm was developed to automatically detect the therapy task a subject performed based on the gyroscope and electromyography data. The system classified which task a subject was attempting to perform with greater than 80% accuracy despite the fact that those with severe impairment produced movements that did not resemble the goal tasks and were visually indistinguishable from different tasks. This developed system could potentially be used for home-therapy compliance monitoring, real-time patient feedback and to control therapy interventions.