Assessing upper limb functional use in daily life using accelerometry: A systematic review

Area of Science:

• Biomechanics and Rehabilitation Engineering
• Wearable Sensor Technology
• Data Science in Healthcare

Background:

• Upper limb dysfunctions significantly impact daily activities and quality of life.
• Accelerometer data are widely utilized to quantify upper limb usage, but analysis methods vary.
• Standardized assessment and data analysis are crucial for accurate classification of upper limb function.

Approach:

• Conducted a systematic literature review across major scientific databases (PubMed, Embase, Scopus, Web of Science, etc.).
• Included studies reporting accuracy and/or validity of accelerometer-based methods for upper limb functional use assessment.
• Focused search terms on "upper limb," "activity tracking," and "functional activity."

Key Points:

• Reviewed 13 studies employing counts threshold, gross movement scores, and machine learning models.
• Machine learning models demonstrated high classification accuracy (68-97% intrasubject, 59-92% intersubject).
• Both machine learning and counts threshold methods showed high validity and accuracy for assessing upper limb function.

Conclusions:

• Accelerometer-based methods, particularly machine learning, offer accurate and valid assessment of upper limb functional use.
• These objective measures can significantly enhance the evaluation of upper limb dysfunctions in clinical settings.
• Accelerometry provides valuable, reliable data for understanding and managing upper limb impairments.