Artificial Intelligence for skeleton-based physical rehabilitation action evaluation: A systematic review

Area of Science:

• Rehabilitation Engineering
• Computer Vision
• Artificial Intelligence

Background:

• Home-based rehabilitation programs are crucial for individuals with physical disabilities to regain strength and balance.
• Patients often lack expert supervision, hindering accurate self-assessment of exercise performance.
• Advancements in vision-based sensors and AI offer potential for automated monitoring solutions.

Approach:

• This paper reviews literature on skeleton data acquisition for physiotherapy exercise monitoring.
• It examines Artificial Intelligence (AI) methodologies for analyzing skeleton data, including feature learning.
• The review covers evaluation metrics and feedback generation for rehabilitation monitoring.

Key Points:

• Vision-based sensors capture accurate skeleton data, enabling detailed activity monitoring.
• Computer Vision (CV) and Deep Learning (DL) advancements facilitate automatic patient monitoring.
• AI-based analysis of skeleton data is key for feature learning, evaluation, and feedback in rehabilitation.

Conclusions:

• The study provides a comprehensive review of AI-driven approaches for physiotherapy exercise monitoring.
• It highlights challenges in skeleton data acquisition, analysis, and feedback generation.
• Future research directions are proposed to advance AI applications in home-based rehabilitation.