Related Experiment Video
Updated: Jul 8, 2025
08:56
Clinical Assessment of Spatiotemporal Gait Parameters in Patients and Older Adults
Published on: November 7, 2014
13.9K
Image-based Gait Spatiotemporal Parameters Estimation using a Single Camera and CNN-Transformer Hybrid Network.
Summary
This study introduces a new vision-based method for detailed stride-by-stride gait analysis in the elderly. The approach accurately estimates gait parameters, aiding remote health monitoring and early disease detection.
Area of Science:
- Biomedical Engineering
- Gerontology
- Computer Vision
Background:
- Remote health monitoring of the elderly is crucial for timely intervention.
- Existing vision-based gait analysis often provides only average parameters, limiting detailed assessment.
- Accurate, continuous gait parameter estimation is needed for effective elderly care.
Purpose of the Study:
- To develop a straightforward, vision-based method for estimating stride-by-stride gait spatiotemporal parameters.
- To enable remote and continuous health monitoring of elderly individuals through gait analysis.
- To validate the proposed method against a gold standard system.
Main Methods:
- Trained three deep learning networks using RGB frames from a mobile camera.
- Input: few RGB frames; Output: continuous 1D signal of spatial and temporal gait parameters.
- Simultaneously collected data using a GAITRite system and a mobile camera from 160 elderly individuals.
Main Results:
- Estimated stride lengths with high correlations (0.938+).
- Detected gait events with excellent F1-scores (0.914+).
- Demonstrated excellent agreement with the GAITRite system for spatiotemporal gait parameter analysis.
Conclusions:
- The proposed vision-based method accurately estimates stride-by-stride gait parameters.
- This approach facilitates remote, continuous health monitoring of the elderly.
- It supports early disease diagnosis, treatment, and intervention through gait analysis.
