A Wearable Soft Robot for Stroke Patients’ Finger Occupational Therapy and Quantitative Measures on the Joint Paralysis

Abstract

We suggest a wearable soft robot for post-stroke, hemiplegic patients’ finger rehabilitation and quantitative evaluations on the joint paralysis. The device consists of a pair of gloves. One measures finger positions of the normal side, and the other induces symmetric movements on the affected side by pneumatic force. Ten patients at Brunnstrom stage 3 and 4 from a local hospital participated in this study. They performed Rolyan Stacking Cones 10 times with and without the support of the soft robot. We measured subjects’ proximal interphalangeal angles and air pressures in the pneumatic glove during the exercise to monitor grab/release patterns. The soft robot helped open their paralyzed finger joints by more than 50 degrees on average regardless of Brunnstrom stages. We applied pattern recognition methods on the measurement to quantitatively evaluate the subjects. A support vector machine revealed a misclassification rate of 20%, implying that there were a considerable number of overlapping data sets near the boundary between Brunnstrom stages. K-means method with three clusters suggested a new subject group near the support vector machine border. Thus, we conclude that our wearable soft robot not only provides grab/release guides to post-stroke patients but also provides quantitative information on their finger paralysis supplementary to the existing qualitative assessments.