Abstract | ||
---|---|---|
Gait analysis aims to study human motion and its potential association with chronic diseases, such as Parkinson's disease and hemiplegic paralysis, by extracting various gait characteristics. It has been a challenging problem to accurately extract temporal and spatial gait parameter and to explore the relationship between gait signal and a disease of interest. In this study, we introduce a gait sensing platform that can capture human movement and classify patients with Parkinson's disease from healthy subjects. Specifically, we first show the platform that consists of force sensitive pressure sensors. Second, we extract gait features from the gait signal collected from the platform. Finally, we collect experimental data from 386 volunteers, including 218 healthy subjects and 168 patients with Parkinson's disease, and conduct extensive experiments to show the possibility of classifying Parkinson's disease patients at a high confidence level. Experimental results of nine different classifiers show that the random forest model outperforms the other eight competitors and obtains an accuracy of 92.49%, demonstrating the power of quantitative gait analysis in the early detection of Parkinson's disease. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1109/BIBM.2017.8218048 | 2017 IEEE INTERNATIONAL CONFERENCE ON BIOINFORMATICS AND BIOMEDICINE (BIBM) |
Keywords | DocType | ISSN |
Gait cycle, Gait parameter, Gait analysis, Parkinson's disease, Random forest | Conference | 2156-1125 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
6 |