Circadian Rhythm Stability Analysis from Actigraphy Data in Persons with Dementia

Circadian rhythms strongly influence psychiatric disorders such as dementia. The stability of circadian rhythms is of high interest in medical research and practice, yet the stability in the dynamic sense remains unexplored. In this study we introduce a set of indicators based on stability analysis. Actigraphy data collected from persons with dementia over seven days, four months apart, is filtered, and then frequency-based model fitting is performed, to which both pole-placement and damping factor analysis is applied. Concurrently, a method based on a multi-harmonic sine model is designed in collaboration with clinical experts to obtain an at-a-glance visualization of circadian rhythms. The method is scalable for multiresolution applications. Results show the capabilities of dynamic stability analysis based on actigraphy data.