A Multi-Level EEG–EMG Neurofeedback Platform for Hand Rehabilitation After Stroke

Hand rehabilitation in neurologic conditions such as stroke and cerebral palsy traditionally emphasizes repetitive task practice with visually observable feedback, despite motor impairment arising largely from abnormal neuromuscular activation. We present a platform that leverages noninvasive measurements of brain and muscle activity for neurofeedback-guided movement training. Trainees first learn to control EEG during movement preparation, followed by reciprocal control of finger muscle EMG during exoskeleton-assisted movement. We describe the platform design and two feasibility studies. Five neurotypical individuals learned to use EEG and EMG to drive an exoskeleton to grasp and release a virtual ball in a single session. They achieved a mean success rate of 65%, demonstrating improved movement latency (9%) and task completion time (6%) across the session. One individual post-stroke trained with the platform across eight sessions and exhibited improvements on the Box and Blocks Test, the Action Research Arm Test, and the Wolf Motor Function Test. These results demonstrate the feasibility of multi-level, neurofeedback training that targets neural activation throughout movement, rather than movement outcome alone. By explicitly engaging both cortical and muscular control signals, this paradigm offers a promising new direction for hand rehabilitation following neurologic injury.