Modeling Wearable Data for Predicting Individual Circadian Rhythms

Abstract

The study of circadian rhythms and the epidemiology of circadian disruption is an emerging area of interest for the fields of chronobiology and chrono-medicine; however, validated measurements of circadian markers have mostly been confined to laboratory settings. The main aim of this project is to explore and systematically map the methods of monitoring circadian markers through wearable devices and review the biologically inspired mathematical models used for the prediction of circadian rhythms, so that they can be applied in general population [1]. The ultimate goal is to find more time and cost-effective methods of tracking compared to the in-lab circadian gold standard markers (core body temperature and plasma levels of cortisol and melatonin).

The project consists of an overview of the human circadian rhythms, followed by the presentation of physiological parameters that have been used for estimation (both inside and outside of the laboratory). The project further outlines the basic principles of mathematical modeling of circadian rhythms [2] and estimates which parameters and models are the best fit for wearable consumer-grade devices, such as the Apple Watch. Our main preliminary finding is that movement as a sole parameter could be used as a proxy for light in such cases [1]. Another promising approach is statistical modeling that extracts underlying circadian rhythmicity from human heart rate [3].

In conclusion, the project estimates the outcomes and performance of such models and presents some possible future applications. These include overcoming disruptions due to phenomena such as sleep-wake disorders, (social) jet lag, shift work, etc., which have already been implemented in some mobile phone apps. Regarding gaps in research, since this is a fairly new area of study, the main aim is to improve the predictability of models for more complex situations (e.g. rotating shift work). Furthermore, the project also highlights the importance of these approaches for future translational studies and the field of personalized medicine, enabling a more precise timing of administration of various therapies.

References

[1] Y. Huang et al., “Predicting circadian phase across populations: a comparison of mathematical models and wearable devices,” Sleep, vol. 44, no. 10, May 2021. doi:10.1093/sleep/zsab126

[2] A. Asgari‐Targhi and E. B. Klerman, “Mathematical modeling of circadian rhythms,” WIREs Systems Biology and Medicine, vol. 11, no. 2, Oct. 2018. doi:10.1002/wsbm.1439

‌[3] C. Bowman et al., “A method for characterizing daily physiology from widely used wearables,” Cell Reports Methods, vol. 1, no. 4, p. 100058, Aug. 2021. doi:10.1016/j.crmeth.2021.100058