Deep learning-based physical exercise assessment of older adults using single-camera videos

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Geriatric Rehabilitation Technology

Background:

• Older adults in care homes often lack supervised physical activity, hindering functional independence.
• Technology-supported exercise platforms require automated monitoring for safety and quality assurance.
• Personalized exercise supervision is crucial but resource-intensive in residential care settings.

Purpose of the Study:

• To design and evaluate a deep learning (DL) system for automated exercise recognition and performance assessment.
• To enable unsupervised, technology-supported exercise monitoring in care homes using standard video recordings.
• To estimate objective exercise performance metrics (EPMs) for personalized rehabilitation feedback.

Main Methods:

• Developed a DL model using temporal convolutional neural networks to process 2D skeleton poses from video.
• The system recognizes exercise types and estimates 3D joint angles for EPM calculation (duration, repetitions, motion variability, range of motion).
• Validated the system with seven care-home residents using inertial sensors for ground-truth joint angles and EPMs.

Main Results:

• The DL model achieved high accuracy in exercise recognition (F1@50 of 0.92) and joint angle estimation (MPJAE of 7.7°).
• Estimated EPMs showed strong agreement with ground truth, with high correlations for duration (0.93) and repetition count (0.86).
• The system reliably estimated motion variability and range of motion across various rehabilitation exercises.

Conclusions:

• A DL-based video monitoring system can reliably assess exercise performance in older adults.
• This technology facilitates unsupervised, safe, and quality-assured exercise in residential care settings.
• The developed pipeline offers a scalable solution to support physical activity and functional independence in geriatric populations.