Fig. 3: The relationship between daily mood on the current-day and circadian disruption measures on the subsequent-day.

a, b The correlation between current day mood score and CRCO-sleep misalignment level on the following day before the internship (a) and during the internship (b). c, d The correlation between current day mood score and CRPO-sleep misalignment level on the following day before the internship (c) and during the internship (d). e, f The correlation between current day mood score and internal misalignment level on the following day before the internship (e), and during the internship (f). The next day circadian misalignment levels were binned by the corresponding mood score. Bin edges were computed every 0.5, and the mean and standard error of the mean of misalignment levels were plotted against the midpoint of each bin. The line and shaded region represent the best-fitted line and the corresponding 95% confidence band, respectively, obtained by applying linear least squares regression to the binned data. b, ρ, and p denote the beta estimate, Pearson’s correlation coefficient, and p-value, respectively. g, h The negative correlation between current day mood score and circadian disruption measures on the following day persisted even after adjustment for demographic and geographic variables using a generalized estimating equation analysis (g) and a linear mixed-effects modeling (h). The 95% confidence intervals of the estimated beta values and their respective p-values are indicated with error bars and asterisks. A two-tailed Wald t-test was used to compute p-values and assess statistical significance. *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001.