Using Contingent Reinforcement to Augment Muscle Activation After Perinatal Brachial Plexus Injury: A Pilot Study.

AIM: Examine the feasibility of increasing muscle activation with electromyographically (EMG)-triggered musical-video as reinforcement for children with perinatal brachial plexus injury (PBPI). METHODS: Six children with PBPI (9.3 ± 6.3 months; 5 female, 1 male) and 13 typically developing (TD) controls (7.8 ± 3.5 months; 4 female, 9 males) participated. The left arm was affected in 5/6 children with PBPI. We recorded the integral (Vs) of biceps activation with surface EMG during two conditions per arm in one session: (1) 100 second (s) baseline without reinforcement and (2) 300 s reinforcement (musical-video triggered to play with biceps activation above threshold [V]). We examined the relation between the mean integral with reinforcement and hand preference. RESULTS: Mean biceps activation significantly increased from baseline in the affected arm of the group with PBPI by the 2nd (p < .008) and 3rd (p < .0004) 100 s intervals of reinforcement. Six of 6 children with PBPI and 12/13 TD controls increased activation in at least one arm. A lower integral was linked with hand preference for the unaffected right side in the PBPI group. CONCLUSION: This study supports contingent reinforcement as a feasible method to increase muscle activation. Future work will examine training dose and intensity to increase arm function.

Human elbow joint torque is linearly encoded in electromyographic signals from multiple muscles.

When the central nervous system (CNS) develops a muscular activation pattern to accomplish a particular isometric task, it clearly uses information concerning the external task requirements. These task requirements serve as inputs to neural transformations that output muscular activations. However, the nature of the inputs is not exactly known. Electromyographic (EMG) signals from eight muscles spanning the human elbow, as well as the total joint torque, were collected during a submaximal isometric flexion/extension task at a single joint angle. The EMG data, without any torque information, were subjected to principal components analysis. We found that 98% of EMG data variation could be described by two principal components the first resembled the joint torque and the second resembled the sum of the EMG signals from all eight muscles. The findings suggest that the CNS encodes these two quantities during isometric tasks.