Complexity Analysis of EMG Signals for Patients After Stroke During Robot-Aided Rehabilitation Training Using Fuzzy Approximate Entropy

The paper presents a novel viewpoint to monitor the motor function improvement during a robot-aided rehabilitation training. Eight chronic poststroke subjects were recruited to attend the 20-session training, and in each session, subjects were asked to perform voluntary movements of elbow flexion and extension together with the robotic system. The robotic system was continuously controlled by the electromyographic (EMG) signal from the affected triceps. Fuzzy approximate entropy (fApEn) was applied to investigate the complexity of the EMG segment, and maximum voluntary contraction (MVC) during elbow flexion and extension was applied to reflect force generating capacity of the affected muscles. The results showed that the group mean fApEn of EMG signals from triceps and biceps increased significantly after the robot-aided rehabilitation training (p < 0.05). There was also significant increase in maximum voluntary flexion and extension torques after the robot-aided rehabilitation training (p < 0.05). There was significant correlation between fApEn of agonist and MVC (p < 0.01), which implied that the increase of motorneuron number is one of factors that may explain the increase in muscle strength. These findings based on fApEn of the EMG signals expand the existing interpretation of training-induced function improvement in patients after stroke, and help us to understand the neurological change induced by the robot-aided rehabilitation training.