Detecting Walking Challenges in Gait Patterns Using a Capacitive Sensor Floor and Recurrent Neural Networks.

Raoul Hoffmann,Marcin Grzegorzek,Axel Steinhage,Hanna Brodowski

doi:10.3390/s21041086

Raoul Hoffmann, Marcin Grzegorzek + Show 2 more

Open Access

https://doi.org/10.3390/s21041086

Copy DOI

Abstract

Gait patterns are a result of the complex kinematics that enable human two-legged locomotion, and they can reveal a lot about a person’s state and health. Analysing them is useful for researchers to get new insights into the course of diseases, and for physicians to track the progress after healing from injuries. When a person walks and is interfered with in any way, the resulting disturbance can show up and be found in the gait patterns. This paper describes an experimental setup for capturing gait patterns with a capacitive sensor floor, which can detect the time and position of foot contacts on the floor. With this setup, a dataset was recorded where 42 participants walked over a sensor floor in different modes, inter alia, normal pace, closed eyes, and dual-task. A recurrent neural network based on Long Short-Term Memory units was trained and evaluated for the classification task of recognising the walking mode solely from the floor sensor data. Furthermore, participants were asked to do the Unilateral Heel-Rise Test, and their gait was recorded before and after doing the test. Another neural network instance was trained to predict the number of repetitions participants were able to do on the test. As the results of the classification tasks turned out to be promising, the combination of this sensor floor and the recurrent neural network architecture seems like a good system for further investigation leading to applications in health and care.

Highlights

Evaluating gait patterns is an essential resource in diagnosing neurological states or orthopedic problems
One can use sensors that are below the floor and which deliver the time and location of steps—the specific sensor floor model that was used for this study, SensFloor®, achieves this by detecting changes in the electric capacitance on a grid of sensor fields
The main contribution of this paper is in the methodological setup, which is a combination of using the floor sensor hardware for recording gait patterns, processing the raw data with its unique properties, and the application of machine learning models for the analysis tasks on the specific kind of data that was gathered

Summary

Introduction

Evaluating gait patterns is an essential resource in diagnosing neurological states or orthopedic problems. Gait can be analysed by formulating a task of interest as a classification or regression problem and train a machine learning model, which is the approach we chose for this project This is useful when working with sensors, where patterns can be found directly in the data stream without a detour of calculating intermediate, semantically meaningful or descriptive parameters or features. To induce changes in the gait patterns, extra requirements were set for the participants to be fulfilled while walking, like putting on a blindfold or performing a dual-task like spelling backwards [4,5] These tasks were chosen as it was expected that they generate the very small variations in the gait patterns which were needed for the system evaluation. A trained system that automatically delivers relevant hints or predicts parameters which are of high clinical interest would save a lot of time and work in everyday clinical practice as it could help in identifying those patients who may benefit from targeted interventions

Objectives

Methods

Results

Discussion

Conclusion