Related Experiment Video
Updated: Sep 29, 2025
07:24
Home-Based Monitor for Gait and Activity Analysis
Published on: August 8, 2019
6.8K
Explainable Artificial Intelligence and Wearable Sensor-Based Gait Analysis to Identify Patients with Osteopenia and
Jeong-Kyun Kim1,2, Myung-Nam Bae2, Kangbok Lee2
1Department of Computer Software, University of Science and Technology, Daejeon 34113, Korea.
Biosensors
|March 24, 2022
Summary
This study introduces a wearable gait analysis tool to detect risks of osteopenia (bone loss) and sarcopenia (muscle loss) using inertial sensors. The method accurately identifies these conditions, aiding in daily life management for seniors.
Area of Science:
- Gerontology
- Biomedical Engineering
- Musculoskeletal Health
Background:
- Osteopenia and sarcopenia significantly impact seniors' quality of life and are linked to various age-related diseases.
- Current diagnostic methods for osteopenia and sarcopenia often require specialized hospital settings, limiting accessibility for daily monitoring.
- Gait analysis shows promise as a non-invasive indicator for musculoskeletal health, offering potential for portable, real-world assessments.
Purpose of the Study:
- To develop and validate a portable gait analysis method using wearable inertial sensors for assessing osteopenia and sarcopenia risks in daily life.
- To identify key gait parameters indicative of osteopenia and sarcopenia using explainable artificial intelligence (XAI).
- To evaluate the performance of machine learning models in classifying osteopenia and sarcopenia based on gait data.
Main Methods:
- Inertial sensor data from a wearable gait device was collected and classified into seven distinct gait phases.
- Descriptive statistical parameters were extracted for each gait phase to characterize gait patterns.
- Explainable AI techniques, including XGBoost and random forest, were employed to analyze the importance of gait parameters for osteopenia and sarcopenia classification. Transfer learning with ResNet was also explored.
Main Results:
- The XGBoost model achieved 88.69% accuracy in identifying osteopenia risk.
- The random forest model demonstrated high accuracy (93.75%) in identifying sarcopenia risk.
- Gait analysis using descriptive statistical parameters and AI models proved more accurate than transfer learning approaches for this specific classification task.
Conclusions:
- Wearable inertial sensor-based gait analysis provides a highly accurate and statistically significant method for managing osteopenia and sarcopenia risks.
- The identified gait factors offer valuable insights for developing personalized interventions and monitoring bone and muscle loss in older adults.
- This technology facilitates convenient, daily-life assessment of age-related musculoskeletal decline, improving proactive health management.
