Gait Segmentation of Data Collected by Instrumented Shoes Using a Recurrent Neural Network Classifier

Antonio Prado; Xiya Cao; Maxime T Robert; Andrew M Gordon; Sunil K Agrawal

doi:10.1016/j.pmr.2018.12.007

Gait Segmentation of Data Collected by Instrumented Shoes Using a Recurrent Neural Network Classifier

Phys Med Rehabil Clin N Am. 2019 May;30(2):355-366. doi: 10.1016/j.pmr.2018.12.007. Epub 2019 Mar 11.

Authors

Antonio Prado¹, Xiya Cao¹, Maxime T Robert², Andrew M Gordon², Sunil K Agrawal³

Affiliations

¹ Mechanical Engineering, Columbia University, 500 West 120th Street, New York, NY 10027, USA.
² Department of Biobehavioral Sciences, Teachers College, Columbia University, 525 West 120th Street, Box 93, New York, NY 10027, USA.
³ Columbia University, 500 West 120th Street, Mail Code: 4703, New York, NY 10027, USA. Electronic address: sunil.agrawal@columbia.edu.

PMID: 30954152
DOI: 10.1016/j.pmr.2018.12.007

Abstract

The authors present a Recurrent Neural Network classifier model that segments the walking data recorded with instrumented footwear. The signals from 3 piezoresistive sensors, a 3-axis accelerometer, and Euler angles are used to generate temporal gait characteristics of a user. The model was tested using a data set collected from 28 adults containing 4198 steps. The mean errors for heel strikes and toe-offs were -5.9 ± 37.1 and 11.4 ± 47.4 milliseconds. These small errors show that the algorithm can be reliably used to segment the gait recordings and to use this segmentation to estimate temporal parameters of the subjects.

Keywords: Gait recognition; Machine learning; Neural network; Rehabilitation robotics; Wearables.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Adolescent
Adult
Cerebral Palsy / diagnosis
Cerebral Palsy / physiopathology
Child
Equipment Design
Female
Gait Analysis / instrumentation*
Gait Analysis / methods*
Humans
Male
Neural Networks, Computer*
Robotics
Shoes*
Young Adult