Enhanced Classification of Abnormal Gait Using BSN and Depth

Abstract

Changes in gait can be caused by a wide range of health complications. As deviations in gait may be an indicator of deteriorating health, abnormalities can be used as a surrogate measure for detecting the onset of certain symptoms. Previous studies have demonstrated the value of wearable sensing for gait analysis. This paper demonstrates the added value of using a depth vision sensor combined with wearable sensors for gait analysis. It also presents a method for extracting a robust set of depth features. The preliminary results from a simulated homecare environment using a three-layer artificial neural network classifier demonstrate the advantages of using a depth sensor for gait analysis.