Gait Analysis for Post-Stroke Hemiparetic Patient by Multi-Features Fusion Method

• DOI: 10.3390/s19071737
• Publication Date: 2019-04-12T07:46:37Z
• Authors: Mengxuan Li; Shanshan Tian; Linlin Sun; Xi Chen
• Keywords: Dynamic Time Warping; Sample entropy; Hemiparesis; Stroke

Walking is a basic requirement for participating in daily activities. Neurological diseases such as stroke can significantly affect one's gait and thereby restrict activities that are part of living. Previous studies have demonstrated temporal parameters useful characterizing post-stroke hemiparetic gait. However, no previous investigated the symmetry, regularity stability gaits. In this study, dynamic time warping (DTW) algorithm, sample entropy method empirical mode decomposition-based index were utilized to obtain three aforementioned types features, respectively. Studies conducted with 15 healthy control subjects survivors. Experimental results revealed proposed features could differentiate patients from by Mann-Whitney test (with p-value less than 0.05). Finally, four representative classifiers order evaluate possible capabilities these distinguish gaits subjects. The maximum area under curve values shown be 0.94 k-nearest-neighbor (kNN) classifier. These promising illustrated considerable potential promote future design automatic analysis systems clinical practice.