Our research focused on the relationship between sleep, medication use, and rhythms. 

Going beyond detection of sleep, we then examined the rhythmicity of the app-counts, which reflects the regular fluctuation of human-smartphone interactions. The well-established method to analyze rest-activity rhythms (RAR) is based on actigraphy. Non-parametric analysis of circadian rhythms, introduced by (Witting et al.) in 1990, uses acti-counts to calculate 3 indicators of circadian rhythm: interdaily stability (IS), relative amplitude (RA), and intradaily variability (IV). These indicators measure various scales of rhythmicity. Specifically, IS represents inter-daily rhythm stability over at least a week, RA looks at the difference of activity level between day and night, and IV focuses on active level stability between each hour within a day.

Built upon daily passive records of physical activities via actigraphy and human-smartphone interactions via mobile phone apps, we showed the disruption of circadian rhythm, measured by non-parametric indexes, is correlated with diseases like depression, obesity, and acute stroke. This correlation sheds light on the indexes’ potential to serve as biomarkers for the diseases, specifically, RAacti and ISacti for post-stroke rehabilitation, ISapp and ISacti for obesity, and IVapp with RAacti for depressive symptoms.

We developed a low-power app called Rhythm that tracks sleep patterns through smartphone usage. It estimates sleep duration and routine stability, offering insights into the effects of pre-sleep phone use and long-term sleep behavior on health. Rhythm has strong potential for clinical use in sleep medicine and psychiatry.