Body activity grading strategy for cervical rehabilitation training

Area of Science:

• Rehabilitation Medicine
• Biomedical Engineering
• Computer Science

Background:

• Cervical rehabilitation requires accurate exercise assessment for effective patient recovery.
• Current methods for evaluating exercise performance can be subjective and time-consuming.
• Computer-assisted systems offer potential for objective and efficient rehabilitation monitoring.

Purpose of the Study:

• To develop and validate a novel body activity grading strategy for computer-assisted cervical rehabilitation.
• To utilize a hidden Markov model for exercise phase segmentation and kinematic feature scoring.
• To assess the accuracy of the proposed grading strategy against manual evaluations by physical therapists.

Main Methods:

• A hidden Markov model was employed to segment cervical rehabilitation exercises into distinct phases.
• A scoring reference was developed to evaluate kinematic features within each identified phase.
• The proposed strategy was applied to 34 cervical rehabilitation exercises and compared with manual assessments.

Main Results:

• The proposed method achieved an average phase segmentation difference of 93 ms compared to manual methods.
• The average phase scoring difference between the proposed and manual approaches was 0.045.
• The overall grading difference for the evaluated samples was 5.5%, indicating high accuracy.

Conclusions:

• The proposed body activity grading strategy demonstrates accuracy comparable to that of physical therapists.
• The method is suitable for providing online supervision during computer-assisted cervical rehabilitation training.
• This technology has the potential to enhance the effectiveness and accessibility of rehabilitation programs.