Automatic Functional Shoulder Task Identification and Sub-task Segmentation Using Wearable Inertial Measurement Units...

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Clinical Biomechanics

Background:

• Current frozen shoulder assessment tools rely on manual operation, introducing bias and data labeling challenges.
• Objective, continuous, and quantitative data are crucial for reliable shoulder assessment.
• Advanced sensor technologies offer potential for improved diagnostic tools.

Purpose of the Study:

• To develop an automated system for functional shoulder task identification and sub-task segmentation using inertial measurement units (IMUs).
• To overcome limitations of manual assessment, providing reliable task labeling and sub-task information.
• To enhance the accuracy and usability of sensor-based frozen shoulder evaluation tools.

Main Methods:

• A hierarchical system combining machine learning models and rule-based modifications for task identification and sub-task segmentation.
• Utilized inertial measurement units (IMUs) to capture shoulder movement data.
• Evaluated the system on nine healthy subjects and nine frozen shoulder patients performing common shoulder tasks.

Main Results:

• Achieved an 87.11% F-score for shoulder task identification.
• Attained an 83.23% F-score for sub-task segmentation.
• Demonstrated a mean absolute time error of 427 milliseconds for sub-task segmentation.

Conclusions:

• The proposed automated system effectively identifies shoulder tasks and segments sub-tasks with high accuracy.
• This approach offers a feasible solution for reliable and objective frozen shoulder assessment in clinical settings.
• The method has the potential to improve the evaluation process for clinical professionals.