Related Experiment Video
Updated: Aug 9, 2025
07:24
Home-Based Monitor for Gait and Activity Analysis
Published on: August 8, 2019
6.8K
Open-source dataset reveals relationship between walking bout duration and fall risk classification performance in
Brett M Meyer1,2, Lindsey J Tulipani3, Reed D Gurchiek4
1Department of Electrical and Biomedical Engineering, University of Vermont, Burlington, Vermont, United States of America.
PLOS Digital Health
|February 22, 2023
Summary
Remote monitoring using wearable sensors can identify fall risk in people with multiple sclerosis (PwMS). Longer free-living walking data better distinguishes fallers from non-fallers, with aggregated data yielding the best fall risk classification performance.
Area of Science:
- Neurology
- Biomedical Engineering
- Rehabilitation Science
Background:
- Falls are a significant cause of morbidity in people with multiple sclerosis (PwMS).
- Disease symptoms fluctuate, and standard clinical visits are insufficient for capturing real-time variability.
- Wearable sensors offer a promising approach for remote monitoring of disease progression and fall risk.
Purpose of the Study:
- To introduce a new open-source dataset for investigating fall risk and daily activity in PwMS using remote monitoring.
- To explore the utility of free-living walking data for characterizing fall risk in PwMS.
- To compare free-living data with controlled laboratory data and assess the impact of walking bout duration.
Main Methods:
- Collected inertial-measurement-unit (IMU) data from 38 PwMS (21 fallers, 17 non-fallers) across 11 body locations in lab and free-living settings.
- Acquired patient-reported surveys and neurological assessments.
- Analyzed gait parameters and fall risk using feature-based and deep learning models, varying walking bout durations.
Main Results:
- Gait parameters and fall risk classification performance varied significantly with walking bout duration.
- Deep learning models outperformed feature-based models on free-living data.
- Longer free-living walking bouts showed greater differences between fallers and non-fallers, with aggregated data providing the highest classification performance.
Conclusions:
- Free-living walking data, particularly longer bouts, are valuable for assessing fall risk in PwMS.
- Wearable sensor data from daily life offer a more sensitive measure of fall risk than controlled lab settings.
- Aggregating all free-living walking data provides the most robust performance for fall risk classification in PwMS.
