Distinct Circadian Assessments From Wearable Data Reveal Social Distancing Promoted Internal Desynchrony Between Circadian Markers

Abstract

Mobile measures of human circadian rhythms (CR) are needed in the age of chronotherapy. Two wearable measures of CR have recently been validated: one that uses heart rate to extract circadian rhythms that originate in the sinoatrial node of the heart, and another that uses activity to predict the laboratory gold standard and central circadian pacemaker marker, dim light melatonin onset (DLMO). We first find that the heart rate markers of normal real-world individuals align with laboratory DLMO measurements when we account for heart rate phase error. Next, we expand upon previous work that has examined sleep patterns or chronotypes during the COVID-19 lockdown by studying the effects of social distancing on circadian rhythms. In particular, using data collected from the Social Rhythms app, a mobile application where individuals upload their wearable data and receive reports on their circadian rhythms, we compared the two circadian phase estimates before and after social distancing. Interestingly, we found that the lockdown had different effects on the two ambulatory measurements. Before the lockdown, the two measures aligned, as predicted by laboratory data. After the lockdown, when circadian timekeeping signals were blunted, these measures diverged in 70% of subjects (with circadian rhythms in heart rate, or CRHR, becoming delayed). Thus, while either approach can measure circadian rhythms, both are needed to understand internal desynchrony. We also argue that interventions may be needed in future lockdowns to better align separate circadian rhythms in the body.

Efficient assessment of real-world dynamics of circadian rhythms in heart rate and body temperature from wearable data

“…We found that the large range of the differences greater than 2 h is due to the inter- and intra-individual difference in the relationship between the HR phase and sleep (figure 5 b – d ; electronic supplementary material, figures S6C and D). Our findings that are consistent with the previous results [14,22] indicate that circadian assessment can be performed efficiently using our method.…”

“…The HR model (equation (2.5)) directly accounts for the effects of activity on HR, and indirectly accounts for other changes to HR on shorter timescales through the autocorrelation. Previous work has shown the success of this model in distinguishing HR phase estimates from sleep- or activity-derived timing metrics, and generating parameter estimates that match those from a constant routine [14,22]. The parameters estimated from our new algorithm are also comparable to those from this previous work.…”

“…This method has been tested against HR data from a constant-routine protocol, and HR phase estimates from this method display different dynamics than sleep or activity-derived timings. The HR phase estimates generally aligned with DLMO predictions during normally entrained scenarios, but became desynchronized during social distancing [22]. Together, these findings suggest an ability to capture endogenous circadian signals from algorithms applied to wearable data.…”

“…Here, we proposed a framework allowing efficient estimation of parameters of any regression models with correlated noise. We used this approach to analyse wearable data by adopting the harmonic-regression-plus-first-order-autoregressive models as in the previous studies [14,15,22,30]. Brown and Czeisler showed that the first-order autoregressive noise process is enough to analyse core BT data collected in laboratory settings [15], which matches our findings (figures 3 and 4; electronic supplementary material, figures S3 and S5, and table 1; electronic supplementary material, table S5).…”

“…The human HR shows a circadian variation [17]. This HR circadian signal has also been successfully analysed with a harmonic-regression-plus-first-order-autoregressive model [14,22,30], as in the analysis of the BT rhythm. One difference between the HR model and the BT model is that the HR model contains a term describing the increase in HR by activity, specified as follows:yt=st+d⋅at+vt,where a t is the activity level (i.e.…”