Frailty Level Classification of the Community Elderly Using Microsoft Kinect-Based Skeleton Pose: A Machine Learning...

Area of Science:

• Gerontology
• Biomedical Engineering
• Computer Science

Background:

• Frailty is a critical geriatric syndrome linked to higher risks of disability and hospitalization.
• Early identification of frailty is essential for timely clinical interventions.
• Current assessment methods may not be suitable for real-time, large-scale screening.

Purpose of the Study:

• To develop and validate a machine learning model for real-time classification of elderly frailty levels.
• To utilize skeletal movement data captured by Kinect sensors for frailty prediction.
• To assess the efficacy of different machine learning algorithms in predicting frailty.

Main Methods:

• Recruited 787 community-dwelling elderly individuals.
• Collected skeletal data using Kinect sensors during functional assessments (e.g., arm curl, sit-to-stand, gait analysis).
• Employed machine learning algorithms, including Support Vector Classifier (SVC) and Multi-layer Perceptron (MLP), for frailty prediction.

Main Results:

• The methodology achieved high accuracy in gender classification (up to 84%).
• SVC and MLP models demonstrated exceptional performance in predicting Fried's frailty level, with median accuracies reaching 97.5%.
• The study confirmed the feasibility of using real-time skeletal movement analysis for frailty assessment.

Conclusions:

• Machine learning models, particularly SVC and MLP, can effectively predict elderly frailty levels using Kinect-based skeletal data.
• This approach offers a promising tool for developing real-time clinical predictive assessment systems for geriatric care.
• The findings highlight the potential of leveraging sensor technology and AI for proactive health management in the elderly population.