Algorithmic assessment of shoulder function using smartphone video capture and machine learning.

Area of Science:

• Biomechanics
• Orthopedics
• Medical Imaging

Background:

• Rotator cuff (RC) tears are a leading cause of shoulder pain, particularly in older adults.
• Diagnosis often relies on expensive imaging techniques.
• Previous research has not explored markerless motion capture for assessing shoulder kinematics in RC tears.

Purpose of the Study:

• To investigate the utility of a simple string-pulling task for assessing shoulder mobility in the presence of rotator cuff tears.
• To establish a reliable method for quantifying shoulder kinematics in both animal models and human patients.
• To develop a diagnostic tool for rotator cuff tears using motion analysis.

Main Methods:

• Markerless motion capture technology was employed to analyze shoulder kinematics during a string-pulling task in mice and humans.
• Movement amplitude, duration, and waveform shape were quantitatively assessed.
• A logistic regression model was developed using an ensemble of biomarkers to classify rotator cuff tears.

Main Results:

• Both mice and humans with rotator cuff tears demonstrated reduced movement amplitude and prolonged movement time during the string-pulling task.
• Quantitative changes in movement waveform shape were observed in individuals with rotator cuff tears.
• The developed logistic regression model achieved over 90% accuracy in classifying human patients with rotator cuff tears.

Conclusions:

• A simple string-pulling task, analyzed via markerless motion capture, provides a reliable method for assessing shoulder mobility and detecting rotator cuff tears.
• This approach bridges animal models and human studies, offering a framework for developing novel, accessible diagnostic tools.
• The findings support the potential for smartphone-based, at-home diagnostic tests for shoulder injuries.