Cloud-based rehabilitation and recovery prediction system for stroke patients

Abstract

Motivation is one of the main reasons that a group of patients with the same pathology respond differently to a rehabilitation program. It is well-known that besides family and close friends, showing the real progress to the patients is another significant motivating factor for them. This can increase patient's self-satisfaction, engagement, and enjoyment. In this study, we have designed and implemented a cloud-based rehabilitation system that helps stroke patients enhancing their motor functions. Forty five healthy persons (18 females and 27 males) and three stroke patients have volunteered to participate in our experiment. The derived data from the healthy subjects is served as motion normative data that can be used for accurate assessment of hand function. We have applied a well-known optical alignment technique, dynamic time warping, to compare the time series kinematics patterns of stroke patients with those of healthy subjects. The prototype of this system is tested on three patients for ten weeks. Results show that patients have improved moving and controlling their upper limbs over the course of training. Such conclusion is also confirmed clinically; the patients have performed the action research arm test under a direct supervision of an orthopedic doctor and a professional physiotherapist. Moreover, recovery predictions with ARIMA models of stroke patients have given encouraging results with percentage of error less than 2.0 % for patient one and patient three, and 10.35 % for patient two.