Reduction of Assessment Time for Stroke-Related Impairments Using Robotic Evaluation

Abstract

Robotic technologies can provide objective, reliable tools for assessing a broad range of sensory, motor and cognitive functions. However, as additional tasks are developed on these platforms, the time necessary to assess a patient increases. In this paper, we present a hierarchical task selection strategy for five tasks that form part of the battery of standard tests performed on the KINARM robotic system. The strategy is built using dependencies derived through three types of analyses: 1) non-linear hierarchical ordering theory is applied to determine the ordering of five tasks; 2) the parameters of all tasks are also ranked using non-linear hierarchical ordering theory; and 3) a modeling technique, fast orthogonal search, is applied to assess the predictive power of each robotic task for the estimation of other task parameters. The inferred hierarchical task selection strategy can lead to a reduction of up to 91% of the time required to assess a patient.