Advertisement
Published in:
2020 | OriginalPaper | Chapter
Visualization of Features in Multivariate Gait Data: Use of a Deep Learning for the Visualization of Body Parts and Their Timing During Gait Training
Authors : Yusuke Osawa, Keiichi Watanuki, Kazunori Kaede, Keiichi Muramatsu
Published in: Advances in Industrial Design
Publisher: Springer International Publishing
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Abstract
In this study, we aimed to examine the usefulness of gait classification and feature visualization based on multivariate data for the development of a gait feedback training system capable of considering the physical differences among the trainees. The multivariate data considered in this study were the joint angles and the ground reaction forces. In addition, all multivariate gait data were labeled as gait “rarely associated with stumbling” or “frequently associated with stumbling”. A convolutional neural network was used to learn the gait features. Furthermore, the feature parts of the multivariate gait data used for classification were visualized on a heat map created using Grad-CAM. As the results indicate, a heatmap is able to show the feature parts of a gait frequently associated with stumbling, through which the trainee can adjust their gait.