Concrete targets for tonight & tomorrow, derived from your own recent Garmin data.
How consistently you hit your plan targets over the last 7 days with data.
For each sleep metric, how one night relates to the next night (calendar-adjacent nights only).
Single-subject & observational — Spearman correlations, descriptive not causal.
Each metric vs itself the next night. Positive = a high night tends to repeat (persistence); negative = a high night is followed by a low one (rebound). ✱ = p<0.05.
% change in the next night's metric after a top-third night vs a bottom-third night. Negative = rebound (body corrects downward). ✱ = p<0.05.
Spearman rho. Cyan = the two move together, purple = they move oppositely. Bold + outline = survives multiple-comparison (FDR) correction; dim = not significant (p≥0.05). Hover a cell for details.
When stress spikes, how long until it settles — and does a stressful day leave a mark on the next?
Two timescales from your own data: minutes (from Garmin's 3-minute stress signal) and days.
Garmin "stress" is an HRV-derived proxy for autonomic arousal, not cortisol. Single-subject & observational.
Every high-stress peak (≥76) aligned at t=0, averaged. Stress climbs, spikes, and the very next reading is already back down — a spike, not a plateau. The dashed line is your settled baseline; shaded band is ±1 SEM.
Minutes from each peak until stress holds at rest/low (≤50) for 30 min. Green = settled within the hour; grey = took longer.
Recovery markers around a high-stress day (day 0), in z-scores vs your personal baseline. Resting HR is inverted so up = better recovery for all three. A shared same-day dip that climbs back toward baseline within a few days — no durable multi-day residue.
Daily event count from primary calendar — counts only, no titles or details. Hover the heatmap above to see steps + meetings together.
How today's behavior tracks with tomorrow's metrics — high vs low days, controlling for obvious confounders and corrected for multiple comparisons (Benjamini–Hochberg, q=0.10). Single-subject & observational: associations, not proof of cause. A card is flagged only if it survives that correction.
Outcomes after high vs low exposure days (median split), with a control-variable partial correlation and a permutation p-value. Confidence reflects Benjamini–Hochberg FDR across these tests; a small n means noisy. Association, not causation.
| Cause | Effect | High Days (mean) | Low Days (mean) | Difference | Partial r | n | Confidence |
|---|
Calculating prediction model...
Simple correlations without controlling for confounders. Use causal analysis above for better inference.
| Factor | → Sleep (next day) | → Sleep (+2 days) | → HRV (next day) | → HRV (+2 days) | → Resting HR (next day) |
|---|
How today's activity patterns affect health metrics 1-3 days later.
| Factor | +1 Day Battery | +2 Day Battery | +3 Day Battery | Best Lag |
|---|
Pearson correlation coefficients (−1 to +1). Cell color shows direction; opacity shows strength.
| Metric | Mean | Median | Std Dev | Min | Max | Range |
|---|---|---|---|---|---|---|
| Loading... | ||||||
The average shape of your day — each metric averaged by hour-of-day across the window set above (widen to 90d / All for stable patterns), so the daily rhythm shows through instead of one day's noise. Combined view shades the hour's min–max range; Weekday vs Weekend overlays the two day types.