Robotic-induced improvement of movements in hemiparetics via an implicit learning technique
J.L. Patton, F. A. Mussa-Ivaldi, W. Z. Rymer

Numerous studies have shown that subjects undergo neuromotor adaptation to robot-generated forces applied to the hand during reaching, resulting in after-effects when the forces are removed. We investigated whether this adaptive process might beneficially alter the disturbed hand trajectories of hemiparetics. Individuals with hemiparesis secondary to a cerebral vascular accident in their middle cerebral artery territory were asked to make movements while holding the handle of planar robotic device. They were asked to make straight movements to targets near the center of reachable workspace in one of 3 directions (-60, 60, and 270 degrees, clockwise relative to the anterior axis and randomly presented). Following 60 trials of familiarization and 30 trials of unperturbed "baseline" motions, robotic forces were randomly introduced once every 4 trials for 298 trials as part of an online algorithm that determined the time-dependent force profile that deviates the subject's hand towards a straight-line trajectory. Subjects then trained on the vector opposite of this force profile for 330 trials. We tested for after-effects in catch trials where robot motors were shut off randomly once in every 8 of 120 movements. Our major finding is that there was improvement in performance as an after-effect of adaptation. Furthermore, after 75 movements without any robotic forces, some subjects partially retained their ability to reach in a straight line. The results indicate that it is feasible to capitalize on the nervous system's natural adaptive ability to alter movement patterns, implying that a robot could provide precise and sustained retraining of an impaired individual while simultaneously providing accurate measurement of progress. Supported by NIH F32HD08658 & R24HD39627.

Society for Neuroscience Meeting, November 2001, San Diego, CA