Digital markers and phenotypes of rest-activity rhythms in people with advanced dementia using real-time location data.

BACKGROUND

Disrupted sleep and circadian rhythms in dementia affect rest-activity patterns and impact quality of life, safety, and caregiver burden. The objective of this study is to extend the application of real-time location system (RTLS) technology from nurse call and elopement prevention systems to measuring digital markers and phenotypes of rest-activity in people with advanced dementia in residential care.

METHODS

Wrist-worn RTLS devices continuously tracked location for up to 16 weeks in 47 participants (21 women, mean age 80.1 years) on a specialized dementia unit. Distance moved in 15-min windows was used to derive digital markers, including parametric and non-parametric features. Panel and mixed-effect models were used to investigate the relationship between clinical assessments and RTLS-based digital markers and phenotypes of rest-activity rhythm.

RESULTS

Higher activity intensity was correlated with increased clinical motor agitation scores, and disrupted rhythmicity and reduced time in bed were associated with difficulty falling sleep and increased nighttime motor agitation. Unsupervised machine learning identified 6 distinct rest-activity phenotypes over 1-week periods. These phenotypes include high time in bed, well-regulated, low stability, severe rhythm disturbance, nighttime active, and a highly active individual. Phenotypes differed by age, cognition, mood disturbance, and functional status. Increased activity intensity, decreased rhythmicity, and less nighttime time in bed were associated with higher motor agitation and sleep disruption.

CONCLUSIONS

RTLS-derived rest-activity markers and phenotypes were associated with changes in clinical assessments over time and may support data-driven, evidence-based dementia care.