The association of upper limb sensorimotor capacity, everyday inpatient behavior, and the effects of neurorehabilitation in persons with multiple sclerosis and stroke: a mixed-design study

Background: Quantifying and monitoring the sensorimotor state of persons with neurological disease by means of wearables in everyday life has been shown to be a promising approach. To date, the impact of physical activity volumes in fixed epoch approaches has been limiting the feasibility of kinematic analyses of everyday life upper limb use. Methods: Using acceleration and angular velocity signals from wrist-worn sensors, we collected data of healthy controls (n = 12) as well as persons with multiple sclerosis (n = 17) or stroke (n = 14) during everyday life during inpatient neurorehabilitation. An activity recognition algorithm was used to avoid physical activity volume dependencies that come with epoch-based approaches. Behavioral kinematics were compared between samples and associated with clinical test performance. Further, changes of sensorimotor capacity and behavioral kinematics during neurorehabilitation (n = 15) were analyzed. Results: Physical activity volume independence was achieved. Persons with neurological disease showed less activities and longer activity durations. Further, a PCA suggested three underlying components, namely: behavior, neurological state, and physical state. Components scores were lower (worse) for persons with neurological disease, except for behavior. However, component scores of persons with neurological disease showed great variability in all dimensions. Changes in sensorimotor capacity were partially associated with changes in behavioral kinematics, but effects of neurorehabilitation were mostly seen in outcomes associated with the physical state component. Conclusions: Persons with neurological disease showed neurological impairments as well as declines in the physical condition, which can to some extent be seen in behavioral kinematics. Neurorehabilitation appeared to rather affect the physical than the neurological state. By the novel approach using an activity recognizer instead of fixed epochs, it was possible show traces of sensorimotor capacity, as assessed by clinical tests, in kinematics of everyday life behavior.