Smartwatches Can Detect Walker and Cane Use in Older Adults

Area of Science:

• Gerontology
• Biomedical Engineering
• Rehabilitation Science

Background:

• Assistive devices like walkers and canes are prescribed to reduce fall risk in older adults.
• Consistent use of assistive devices is often low, and adherence measurement is hindered by self-report bias and cognitive issues.
• Changes in walking patterns while using assistive devices present an opportunity for objective monitoring.

Purpose of the Study:

• To investigate the feasibility of using smartphones and smartwatches with machine-learning algorithms to detect assistive device use in older adults.
• To develop and test algorithms capable of differentiating walking with versus without an assistive device.

Main Methods:

• Fourteen older adults wore an Android smartphone and an Actigraph smartwatch.
• Participants completed mobility tests (6-minute walk, 10-meter walk, Timed Up and Go) with and without their assistive device over five days.
• Accelerometer data was used to train and cross-validate machine-learning algorithms for detecting assistive device use.

Main Results:

• Smartwatch-based classifiers demonstrated high accuracy in detecting assistive device use.
• Customized smartwatch algorithms achieved over 95% accuracy for all participants.
• Non-customized smartwatch systems exceeded 90% accuracy for 10 out of 14 participants, performing better for walker users.

Conclusions:

• Leveraging commercial wearable devices and machine learning offers a viable method for monitoring assistive device adherence.
• This approach can inform the development of scalable public health tools for fall prevention.
• Objective quantification of assistive device use can enable timely and proactive interventions for older adults.