Evaluation of similarities between robotic tasks for reduction of stroke assessment time

Robotic technologies provide accurate, objective, and highly reliable tools for assessment of brain function following stroke. KINARM is an exoskeleton device that uses a number of behavioral tasks to objectively quantify sensorimotor, proprioceptive and cognitive brain function using a battery of behavioral tasks. With a growing number of tasks deployed to more broadly assess different aspects of behavior on the KINARM system, different strategies are required to reduce the overall assessment time. Two specific tasks designed for assessment of visuomotor control, multi-joint coordination and cognitive processes related to attention and inhibition include Object Hit and Object Hit and Avoid tasks. The present study investigates the similarities between these two tasks using a system identification technique known as Fast Orthogonal Search. Results of our study show that all parameters of the Object Hit task can be predicted using Object Hit and Avoid parameters with R values ranging from 0.66 to 0.92, close to inter-rater reliability scores for the Object Hit parameters. We were able to classify stroke from control subjects using the predicted Object Hit task parameters, with similar accuracies as those obtained using the actual parameter values. These findings can be used to shorten the KINARM assessment procedure.