Temporal patterns of measured or self-reported data.

a. Total number across all participants of logged dishes (any food or drink intake) per hour (x-axis) and weekday. b. Mean (solid line) and 50% interquartile range (shaded areas) of the blood glucose levels measured during weekdays (black) or weekends (blue). c. Schematic illustrating the experiment to assess if glucose peaks are more likely to be directly preceded by food intake. The time-window of interest (i.e., in which food intake is expected) is displayed in purple, while the control time-window is displayed in gray. d. Distribution of the percentage of glucose peaks directly (purple) or distantly (control—gray) preceded by food intake. Percentages are computed per participant: for each participant, glucose peaks are identified, food intake in the two time-window is reported as a binary variable, and percentages are computed as the fraction of time-windows with reported food intake. e. Distribution of the times at which participants woke up (upper panel) or fell asleep (lower panel) during weekdays (gray) or weekend (blue). These times are either self-reported by participants when filling the morning questionnaires or obtained from participants’ connected devices (smartwatches) data. f. Distribution of average sleep durations (x-axis, in hours) during weekdays (gray) or weekends (blue).