Using a Minimum Set of Wearable Sensors to Assess Quality of Movement in Stroke Survivors

Abstract

The study herein summarized was focused on the development of a method to derive reliable estimates of the quality of movement of stroke survivors via the analysis of wearable sensor data. Data was collected from 34 subjects while they performed a battery of functional movements that are part of a standard clinical assessment. The quality of movement was assessed using the Functional Ability Scale, a validated clinical scale based on visual observation of movement patterns by a clinical expert. Two wearable sensors were positioned on the stroke-affected wrist and the sternum, respectively. Wearable sensor data was processed to derive data features that were in turn used as input to a regression-based Random Forest algorithm that was trained using a leave-one-subject-out method. The Random Forest algorithm generated estimates of the Functional Ability Scale scores. We found out that the estimates generated via analysis of the wearable sensor data were highly correlated (R2=0.97) with the scores generated by the clinical expert.