@theryangray@WHOOP Three month averages are a good way to take the day to day noise out. Did you log anything alongside it, like bedtime or training? That overlay is usually where the story is.
@PsychiatricNews Same shape in the UK Biobank accelerometer data (Windred et al., Sleep 2024, about 61,000 adults). Regularity predicted all-cause mortality more strongly than hours slept there. Observational too, so it points to a marker rather than a proven lever.