What each metric tells you
HRV (heart rate variability)
HRV, measured as overnight SDNN by the Apple Watch, is a proxy for autonomic nervous system balance. Higher values, compared to your own baseline, generally line up with better recovery. Lower values can reflect physical load, illness, sleep debt, alcohol, stress, or just statistical noise. The number on its own is not very useful, the direction across two weeks is.
Resting heart rate (RHR)
RHR is the most stable of the three. A consistent upward shift over a few days, alongside lower HRV, is one of the more reliable signs that recovery is not keeping up with load. A drift downward over months can reflect aerobic conditioning, but it is slow.
Sleep
What matters most week to week is duration and consistency. Sleep stages are interesting but noisy on the Apple Watch. The report leans on duration, regularity, and how the rest of the metrics responded.
Why trends, not single days
Take any one of these metrics and one bad day. There are too many explanations to choose from. Take seven days, with the same direction repeating, and the picture sharpens. Two weeks of data is the threshold where most patterns become reliable enough to act on, which is why VitalBrief uses 14-day baselines for HRV and resting heart rate, with a longer 238-day window for percentile comparisons against your own normal.
How VitalBrief contextualizes them
The Wellness Score combines two pieces:
- Recovery: daily HRV and RHR values converted to percentiles against your baseline, then aggregated across the week.
- Training: calorie volume relative to your baseline, plus a population reference based on body mass.
The two are combined into a 5 to 100 score, with eight labels (peak week, strong week, moderate week, light week, overload week, rest week, strain week, recovery week). The label is sport-aware and takes confidence into account, so a low-confidence high score gets downgraded automatically.
The narrative report uses the score and its commentary as constraints, so the AI cannot say "great recovery week" when the score says otherwise.
Cold start. If you have less than two weeks of usable data, the score does not run yet. The card shows a "Building your pattern" message until enough history exists to compare against.