Affiliations: [a] School of Computer and Information, Hefei University of Technology, Hefei, China. E-mails: [email protected], [email protected] | [b] Institute of Industry & Equipment Technology, Hefei University of Technology, Hefei, China. E-mails: [email protected], [email protected] | [c] Institute of Intelligent Machines, Chinese Academy of Sciences, Hefei, China. E-mail: [email protected] | [d] Hospital Affiliated to Institute of Neurology, Anhui University of Chinese Medicine, Hefei, China. E-mail: [email protected]
Abstract: Gait analysis is the systematic study of human motion and aims at quantifying the gait characteristics with various temporal and spatial gait parameters. Gait has proven to be linked to certain diseases, such as hemiplegic paralysis and Parkinson’s disease (PD). In this work, we present our designed and developed gait sensing platform that can capture human movement, called uGait, and we further present its application in discriminating patients with Parkinson’s disease from healthy subjects. First, we show the overall system architecture of uGait. Second, we detail how to extract and select meaningful spatial-temporal gait parameters and present an effective wrapper-based feature selection method. Typically, we consider the gait parameters associated with the situation where one walks on a u-shape walkway rather than only on a straight walkway, and we consider the influence of human body height on gait parameters extraction. We then apply our proposed feature selector to identify a discriminant subset of gait features and construct a classifier with better generalization ability for PD detection. Finally, we collect experimental data with uGait and conduct extensive experiments with three different classifiers and three different feature selectors. Experimental results demonstrate power of the quantitative gait analysis in classifying PD patients.
