Assessing Free-Living Postural Sway in Persons With Multiple Sclerosis

Area of Science:

• Neurology
• Biomedical Engineering
• Rehabilitation Science

Background:

• Postural instability is a key concern for people with Multiple Sclerosis (PwMS), impacting disease status and fall risk.
• Current postural sway assessments are typically lab-based using force platforms or accelerometers, limiting accessibility.
• Remote monitoring offers a potential solution for accessible, real-world assessment of postural sway.

Purpose of the Study:

• To evaluate the utility of remote postural sway measures for assessing disease status and fall risk in PwMS.
• To compare the performance of machine learning models trained on lab-based versus remote postural sway data.
• To develop and validate a novel framework for analyzing remote patient monitoring data.

Main Methods:

• Collected remote and lab-based postural sway data from 33 PwMS.
• Developed machine learning models to predict fall risk using both data types.
• Applied a subject-specific k-means clustering approach to refine remote data analysis.

Main Results:

• Remote sway measures differed from lab measures but correlated with patient-reported balance and mobility.
• Lab-based fall risk models achieved an AUC of 0.79; initial remote models achieved 0.51.
• The novel clustering approach improved remote model AUC to 0.74 and enhanced associations with patient-reported outcomes.

Conclusions:

• Remote postural sway assessment, particularly with advanced analysis techniques, holds significant potential for evaluating fall risk and balance impairment in PwMS.
• The developed framework offers a promising approach for analyzing data from remote patient monitoring technologies.
• This method enhances the accessibility and utility of real-world data for managing Multiple Sclerosis.