Related Experiment Video
Updated: Jun 6, 2026
08:05
Design and Analysis for Fall Detection System Simplification
Published on: April 6, 2020
Automatic segmentation of triaxial accelerometry signals for falls risk estimation.
Stephen J Redmond1, Maria Elena Scalzi, Michael R Narayanan
1School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, 2052, Australia.
Summary
This study developed algorithms for automatic segmentation of accelerometry data from a directed-routine (DR) test to assess elderly fall risk. Automatic segmentation showed good agreement but slightly reduced correlation with fall risk compared to manual annotation.
Area of Science:
- Gerontology
- Biomedical Engineering
- Rehabilitation Science
Background:
- Falls in the elderly are a major cause of healthcare costs.
- Existing falls detection technology often lacks a proactive prevention strategy.
- Unsupervised monitoring in free-living environments is needed for falls prevention.
Purpose of the Study:
- To develop and validate algorithms for automatic signal segmentation of accelerometry data.
- To enable unsupervised, home-based falls risk assessment using a directed-routine (DR) test.
- To evaluate the accuracy of automatic segmentation for falls risk estimation.
Main Methods:
- A waist-mounted triaxial accelerometer was used to collect data during a self-administered directed-routine (DR).
- Algorithms were developed for automatic segmentation of accelerometry signals.
- Signals from 68 subjects were manually annotated and compared with automatically segmented values.
- Correlation between segmented features and falls risk was assessed.
Main Results:
- The proposed signal segmentation routines demonstrated good agreement with manual annotations.
- Automatic segmentation resulted in a correlation of 0.73 with falls risk.
- Manual segmentation achieved a higher correlation of 0.81 with falls risk.
Conclusions:
- Automatic segmentation of accelerometry data is feasible for unsupervised falls risk assessment.
- While accurate, automatic segmentation shows a slight reduction in falls risk prediction correlation compared to manual methods.
- Further refinement of algorithms may improve the accuracy of automated falls risk assessment in home environments.
