Article: Multi-Risk-Level Sarcopenia-Prone Screening via Machine Learning Classification of Sit-to-Stand Motion Metrics from Wearable Sensors

TitleMulti-Risk-Level Sarcopenia-Prone Screening via Machine Learning Classification of Sit-to-Stand Motion Metrics from Wearable Sensors
Authors
Keywordsartificial intelligence-based diagnosis
good health and wellbeing
microsensors
physical performance
sarcopenia risk levels
sarcopenia-prone detection
Issue Date17-Feb-2025
PublisherWiley Open Access
Citation
Advanced Intelligent Systems, 2025 How to Cite?
Abstract

Sarcopenia, an age-related syndrome characterized by muscle mass and function loss, significantly impacts the quality of life in older adults. A machine learning approach using micro inertial measurement units (μIMUs) for noninvasive sarcopenia-prone screening through a single sit-to-stand (1STS) test is developed. The study involves 53 older participants (65–84 years), each wearing two IMUs, i.e., one on the thigh and one on the waist. The 1STS motion is divided into four phases and extract 510 features from the collected data. Phase 1 is crucial for distinguishing healthy from sarcopenia-prone participants, while Phase 2 is significant in differentiating risk levels. Key indicators include anterior–posterior and mediolateral movements, particularly along the y-axis and z-axis of the sensors. Five classification algorithms (support vector machine (SVM), K-nearest neighbors (KNN), decision tree, linear discriminant analysis, and multilayer perceptron (MLP)) with selected features are trained. The method achieves 98.32% accuracy using SVM and MLP in distinguishing healthy from sarcopenia-prone participants and 90.44% accuracy using KNN in classifying participants across four risk levels (0–3) based on physical performance severity. These results suggest that the proposed method provides a low-cost, nonspecialist technique for large-scale sarcopenia-prone risk screening and assessment of physical performance severities.


Persistent Identifierhttp://hdl.handle.net/10722/355261
ISSN
2023 Impact Factor: 6.8

 

DC FieldValueLanguage
dc.contributor.authorWang, Keer-
dc.contributor.authorZhang, Hongyu-
dc.contributor.authorCheng, Clio Yuen Man-
dc.contributor.authorChen, Meng-
dc.contributor.authorLai, King Wai Chiu-
dc.contributor.authorOr, Calvin Kalun-
dc.contributor.authorHu, Yong-
dc.contributor.authorVellaisamy, Arul Lenus Roy-
dc.contributor.authorLam, Cindy Lo Kuen-
dc.contributor.authorXi, Ning-
dc.contributor.authorLou, Vivian Weiqun-
dc.contributor.authorLi, Wen Jung-
dc.date.accessioned2025-04-01T00:35:17Z-
dc.date.available2025-04-01T00:35:17Z-
dc.date.issued2025-02-17-
dc.identifier.citationAdvanced Intelligent Systems, 2025-
dc.identifier.issn2640-4567-
dc.identifier.urihttp://hdl.handle.net/10722/355261-
dc.description.abstract<p>Sarcopenia, an age-related syndrome characterized by muscle mass and function loss, significantly impacts the quality of life in older adults. A machine learning approach using micro inertial measurement units (μIMUs) for noninvasive sarcopenia-prone screening through a single sit-to-stand (1STS) test is developed. The study involves 53 older participants (65–84 years), each wearing two IMUs, i.e., one on the thigh and one on the waist. The 1STS motion is divided into four phases and extract 510 features from the collected data. Phase 1 is crucial for distinguishing healthy from sarcopenia-prone participants, while Phase 2 is significant in differentiating risk levels. Key indicators include anterior–posterior and mediolateral movements, particularly along the <em>y</em>-axis and <em>z</em>-axis of the sensors. Five classification algorithms (support vector machine (SVM), K-nearest neighbors (KNN), decision tree, linear discriminant analysis, and multilayer perceptron (MLP)) with selected features are trained. The method achieves 98.32% accuracy using SVM and MLP in distinguishing healthy from sarcopenia-prone participants and 90.44% accuracy using KNN in classifying participants across four risk levels (0–3) based on physical performance severity. These results suggest that the proposed method provides a low-cost, nonspecialist technique for large-scale sarcopenia-prone risk screening and assessment of physical performance severities.</p>-
dc.languageeng-
dc.publisherWiley Open Access-
dc.relation.ispartofAdvanced Intelligent Systems-
dc.rightsThis work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.-
dc.subjectartificial intelligence-based diagnosis-
dc.subjectgood health and wellbeing-
dc.subjectmicrosensors-
dc.subjectphysical performance-
dc.subjectsarcopenia risk levels-
dc.subjectsarcopenia-prone detection-
dc.titleMulti-Risk-Level Sarcopenia-Prone Screening via Machine Learning Classification of Sit-to-Stand Motion Metrics from Wearable Sensors-
dc.typeArticle-
dc.identifier.doi10.1002/aisy.202401120-
dc.identifier.scopuseid_2-s2.0-85218011654-
dc.identifier.eissn2640-4567-
dc.identifier.issnl2640-4567-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats