A Postural Assessment Utilizing Machine Learning Prospectively Identifies Older Adults at a High Risk of Falling.

Area of Science:

• Gerontology
• Biomechanical Engineering
• Machine Learning Applications

Background:

• Falls are a leading cause of accidental death and injury in older adults, incurring significant medical costs.
• Current fall risk assessments have limited effectiveness in identifying at-risk individuals.
• Objective and reliable fall risk assessment is crucial for effective prevention strategies.

Purpose of the Study:

• To evaluate the performance of a commercially available, automated machine learning method for assessing fall risk in older adults.
• To determine the reliability and predictive accuracy of the automated method for future fall events.
• To compare the fall prediction capabilities of this method against current standards.

Main Methods:

• A cohort of 209 older adults from senior living facilities and community centers participated.
• Participants underwent a 60-second balance test on a force-plate platform, capturing center-of-pressure data.
• A machine learning algorithm analyzed center-of-pressure data to generate a postural stability (PS) score; fall events were tracked for 12 months.

Main Results:

• The automated method demonstrated high reliability (ICC=0.78).
• Individuals with high fall risk scores (1-3) were three times more likely to fall within a year compared to low-risk individuals (7-10).
• High-risk individuals were significantly more likely to experience spontaneous falls, with survival analysis indicating a median fall event within 9 months.

Conclusions:

• An automated, easy-to-use balance assessment method reliably predicts fall risk in older adults up to a year in advance.
• Objective identification of high-risk individuals using this technology can facilitate personalized fall prevention interventions.
• This machine learning-based approach offers a promising tool for enhancing geriatric fall risk management.