Feasibility of a neurorehabilitation pipeline and an automated algorithm to select appropriate treatments for upper extremity motor paralysis in individuals with chronic stroke.

Abstract

This study aimed to investigate the feasibility of a neurorehabilitation pipeline and develop an algorithm to automatically select the appropriate treatment for individuals with upper extremity motor paralysis after stroke in the chronic phase. In Experiment 1, eight post-stroke participants in the chronic phase who underwent treatment sustaining two to three phases were assessed before and after treatment. In Experiment 2, a decision tree analysis was performed in which the dependent variable was set as the treatment option determined by a board-certified physiatrist for 95 post-stroke participants; the independent variables were only motor function scores or both motor function scores and electromyogram variables. In Experiment 1, the clinical assessment scores were improved significantly after treatment. Experiment 2 showed that the agreements of the model with only motor function scores as the dependent variable and with motor function scores and electromyogram variables as the dependent variables were 75.8% and 82.1%, respectively. This novel treatment package is feasible for improvement of motor function in post-stroke individuals with severe motor paralysis. The study also established an automated algorithm for selecting appropriate treatments for upper extremity motor paralysis after stroke, identifying standard values of key variables, including electromyography variables.