What’s New in This Version
Count missing days as zero
For sometimes-activities like cycling, workouts, or medications, days with no data were previously left out of correlation calculations entirely. That biased results toward days when the activity happened. You can now choose, per data type, to count missing days as zero days instead. Only days where your other recorded data shows your devices were active get counted, so truly unknown days stay unknown. Expect much larger sample sizes, tighter confidence intervals, new causality results, and the disappearance of some “perfect” correlations that were never real.
Look for the new Missing days section on correlation views, or the toggle in each data type’s filter settings.
Also in this update
Fixed help buttons in the filter view dismissing on first tap. The lag range menu now stays open while you toggle multiple ranges. Faster, more responsive setting toggles.