Towards On-Demand Virtual Physical Therapist: Machine Learning-Based Patient Action Understanding, Assessment and...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Rehabilitation Technology

Background:

• Parkinson's disease (PD) significantly impacts patient mobility and balance, necessitating effective rehabilitation.
• Traditional physical therapy (PT) can be costly and inaccessible for many patients with PD.
• Remote monitoring and personalized training are crucial for managing PD progression.

Purpose of the Study:

• To develop a machine learning-based virtual physical therapist (PT) system for personalized remote training of patients with Parkinson's disease (PD).
• To enhance balance and mobility in PD patients through automated, tailored physical therapy exercises.
• To emulate the behavior of a human PT through automated task recommendation and error identification.

Main Methods:

• Development of a two-phase human action understanding algorithm (TPHAU) to analyze patient movements captured via Kinect sensor.
• Implementation of an automated error identification model to pinpoint movement inaccuracies during physical therapy tasks.
• Training a machine learning model using real patient and PT data for personalized, automated task recommendations.

Main Results:

• The proposed TPHAU algorithm achieved high accuracy in understanding patient actions.
• The error identification model demonstrated high accuracy in detecting movement errors.
• The machine learning-based task recommendation model provided accurate and personalized exercise updates.

Conclusions:

• The virtual PT system effectively automates patient movement analysis, error detection, and personalized task recommendations for Parkinson's disease.
• This technology holds significant potential for providing on-demand virtual care, improving accessibility, and reducing healthcare costs for PD patients and providers.
• The system successfully emulates key functions of a human physical therapist, offering a scalable solution for remote rehabilitation.