A computational framework for constructing interactive feedback for assisting motor learning

New motion capture technologies are allowing detailed, precise and complete monitoring of movement through real-time kinematic analysis. However, a clinically relevant understanding of movement impairment through kinematic analysis requires the development of computational models that integrate clinical expertise in the weighing of the kinematic parameters. The resulting kinematics based measures of movement impairment would further need to be integrated with existing clinical measures of activity disability. This is a challenging process requiring computational solutions that can extract correlations within and between three diverse data sets: human driven assessment of body function, kinematic based assessment of movement impairment and human driven assessment of activity. We propose to identify and characterize different sensorimotor control strategies used by normal individuals and by hemiparetic stroke survivors acquiring a skilled motor task. We will use novel quantitative approaches to further our understanding of how human motor function is coupled to multiple and simultaneous modes of feedback. The experiments rely on a novel interactive tasks environment developed by our team in which subjects are provided with rich auditory and visual feedback of movement variables to drive motor learning. Our proposed research will result in a computational framework for applying virtual information to assist motor learning for complex tasks that require coupling of proprioception, vision audio and haptic cues. We shall use the framework to devise a computational tool to assist with therapy of stroke survivors. This tool will utilize extracted relationships in a pre-clinical setting to generate effective and customized rehabilitation strategies.