Deficit Fields for Stroke Recovery

Stroke survivors exhibit error in both reaching extent and abnormal curvatures of motion. Prior error augmentation techniques multiply error by a constant at each instant during movement. However, magnification of spurious errors may provoke over-compensation. We hypothesize that a deficit-field design, using the statistics of a patient's errors to customize training, will provide optimal augmentation that varies during motion as needed. We will compare the training effects of error deficit-fields with previous methods of error augmentation to improve reaching ability.

Motor deficits manifest in the workspace limitations of joints, i.e. reduced range of motion, uneven extension-flexion, inter-joint coupling, and unwanted synergies. Our work builds upon these ideas by augmenting self-directed movement for training coordination. We found that amplifying augmentation can expand motor exploration and improve skill retention in patients. Using motor exploration patterns from each patient, we will form customized deficit-fields to recover normal joint workspace. We will compare augmentation training that either amplifies or diminishes the observed deficits (Expt-1). We also compare deficit-fields with our prior augmentation methods to determine the added value of increased customization (Expt-2).

Clinicians have recognized the benefits of training on everyday tasks (Hubbard, Parsons et al. 2009), as well as practice with whole-body actions (Boehme 1988; Bohannon 1995). However, typical robotic systems have only a single contact point and cannot drive the multiple joints involved in functional tasks. Visual distortions (e.g. a shift, rotation or stretch) can promote adaptation even without forces. Here we present visual distortion of whole body movement during manual tasks during standing, including reaching, grasping, and object manipulation. We compare the training effects of feedback based on deficit-fields versus practice with normal vision.