Mi Reloj Interno (My endogenous clock): an evidence-based mobile-app to improve sleep and circadian rhythms

ABSTRACT

Introduction: The human circadian clock is daily entrained by both light exposure and daily social schedules, which were severely affected during the pandemic-associated lockdown. In a previous work we found that after one month of lockdown, Argentinian residents exhibited later chronotypes compared with a pre-pandemic situation, despite they slept longer and showed less social jetlag (Leone M.J. et al, Current Biology). In this study, we collected an independent set of local data with the aim to develop an evidence-based mobile app that offers customized recommendations to improve and maintain healthy sleep and circadian rhythms. Materials and methods: Data was collected throughout a phone/website survey between July and September 2020 (n=4460, after 4-6 months into lockdown) in Argentina. The survey included questions about demographic factors, habits, and previously standardized and validated questionnaires (MEQ, MCTQ, PSQI). Data from Buenos Aires city and suburbs (n=3246) was calibrated to match the population distribution and it was used to run the main analyses. The rest of the database was used to validate results. We conducted a cross validation process using linear models, which included a feature selection process to find the most relevant regressors to fit each chronotype and sleep-related variable. For a given age and gender, each model predicted a set of optimal values for the regressors (e.g. sunlight exposure, regular activities) where the dependent variable is maximized (or minimized). Finally, the recommendation system is based on the comparison between optimal and actual values for each predictor, considering the most affected variables. Results: The final calibrated sample (age 41.3±15.5, 67% female) shows late chronotypes (MSFsc 0600±11min, MEQ score 49.42±11.3), low levels of social jetlag (0.99h±1.09) and considerable long sleep duration on weekdays (7.31h±1.43). The regressors which significantly affect at least one variable were light exposure, use of alarm, naps and regular activities (and its timing, i.e. work, study, dinner, other activities) as well as age and gender (and interactions). We found no effects of cohabitation, exercise timing and use of screens. The optimal levels of the selected regressors were used to build the recommendation system (i.e. algorithm) on which the mobile app MiRelojInterno is based (available for both Android and iOS platforms, www.mirelojinterno.org). Conclusions: We developed a mobile app based on local evidence that inquires about habits, chronotype and sleep, returns to its users an overview of their current state -including all variables and predictors- along with customized recommendations with the aim to create awareness and improve and maintain healthy sleep and circadian rhythms depending on the age, gender and habits. Acknowledgements This research project was supported by CONICET and Agencia I+D+i (IP-COVID19-679).