Using symbolic aggregate approximation (SAX) to visualize activity transitions among older adults

Area of Science:

• Biomedical Engineering
• Public Health
• Gerontology

Background:

• Objective physical activity monitoring is crucial for understanding health outcomes.
• Self-reported mobility difficulty is a significant factor affecting daily activity.
• Traditional analysis of accelerometer data may not capture nuanced activity pattern changes.

Purpose of the Study:

• To apply Symbolic Aggregate Approximation (SAX) time-series analysis to accelerometer data.
• To visualize activity patterns stratified by self-reported mobility difficulty.
• To identify differences in activity transitions associated with varying levels of mobility impairment.

Main Methods:

• Utilized accelerometer data from 2393 participants in the National Health and Nutrition Examination Survey (NHANES).
• Applied SAX to convert one-minute epoch accelerometry data into four activity levels.
• Examined normalized activity transition prevalence using intelligent icons, stratified by mobility difficulty.

Main Results:

• Individuals with no mobility difficulty exhibited diverse activity transitions across all levels.
• Higher mobility difficulty was associated with constricted activity transitions, primarily around low activity levels.
• Those unable to perform mobility-related activities showed prolonged low-level activity and limited transitions.

Conclusions:

• SAX provides a sensitive method for detecting subtle changes in activity patterns related to mobility difficulty.
• Activity pattern visualization using SAX highlights differences missed by traditional volumetric analysis.
• Understanding these patterns can inform targeted interventions for individuals with mobility limitations.