A tactile sensing approach in stroke rehabilitation

Abstract

The paper describes an experimental, mechanically simple, tactile sensing solution in the form of a sensing chair for discriminating human motion in a reaching task. This costefficient technical approach was employed for the assessment of selective arm movements in stroke survivors. The sensing system classifies trunk motion in a seated stroke survivor during a goal-directed task where there is direct correlation with the level of severity of arm movement. The system interprets motion mechanically from coupled sensory data transients using artificial neural networks and shows tolerance to patients' sitting posture and performance variability. The accuracy of classification was typically greater than 94% across three categories when applied to a group of stroke survivors of wide-ranging motor abilities. The mechanical simplicity, versatility of approach for use in other classes of movement, and potential low cost of manufacturing provides opportunity to employ the system at clinics and homes for assessment and training.