| Metric | Value |
|---|---|
| CPAP usage hours | 6.00 |
| Session count | 1 |
| AHI | 2.33 |
| RDI | 2.67 |
| Obstructive apneas (OA) | 7 |
| Hypopneas (H) | 6 |
| Central apneas (CA) | 1 |
| Unclassified apneas (UA) | 0 |
| RERA | 2 |
| Pressure_Avg | 7.21 |
| Pressure_Min | 4.00 |
| Pressure_Max | 10.36 |
| Pressure_90th | 9.30 |
| Leak_Avg | 2.77 |
| Leak_Max | 22.80 |
| Leak_90th | 10.80 |
| FL_Avg | 0.0117 |
| FL_P95 | 0.08 |
| FL_Max | 0.47 |
| Min_Above_005 (min) | 27.4 |
| Min_Above_010 (min) | 11.3 |
| Min_Above_020 (min) | 2.5 |
| Frac_NonZero (FL samples) | 18.77% |
| Apple Watch time in bed (min) | 390.9 |
| Apple Watch total sleep (min) | 380.4 |
| Apple Watch awake (min) | 10.5 |
| Apple Watch deep sleep (min / %) | 36.6 / 9.63% |
| Apple Watch REM sleep (min / %) | 96.8 / 25.46% |
| Apple Watch core sleep (min / %) | 246.9 / 64.91% |
| Sleep efficiency (TSD/TIB) | 97.30% |
Trailing windows exclude tonight by construction (they cover the 7 and 30 nights ending 2026-10-02).
| Metric | Tonight | 7d avg | 30d avg | Δ vs 7d | Δ vs 30d | 30d z |
|---|---|---|---|---|---|---|
| AHI | 2.33 | 1.55 | 2.08 | +50.3% | +12.0% | +0.36 |
| OA count | 7.00 | 7.71 | 9.50 | -9.2% | -26.3% | -0.73 |
| H count | 6.00 | 2.29 | 3.07 | +162.0% | +95.4% | +1.40 |
| CA count | 1.00 | 0.14 | 0.70 | +614.3% | +42.9% | +0.30 |
| RERA count | 2.00 | 0.71 | 0.67 | +181.7% | +198.5% | +1.51 |
| Pressure_Avg | 7.21 | 7.77 | 7.52 | -7.2% | -4.1% | -0.71 |
| Pressure_90th | 9.30 | 10.29 | 10.03 | -9.6% | -7.3% | -1.40 |
| Leak_Avg | 2.77 | 1.18 | 0.92 | +134.7% | +201.1% | +2.78 |
| Leak_Max | 22.80 | 22.80 | 20.24 | +0.0% | +12.6% | +0.31 |
| Leak_90th | 10.80 | 4.63 | 2.92 | +133.3% | +269.9% | +2.53 |
| Usage hours | 6.00 | 6.54 | 6.40 | -8.3% | -6.3% | -1.01 |
| FL_P95 | 0.08 | 0.09 | 0.09 | -12.1% | -14.0% | -0.75 |
| Min_Above_010 | 11.30 | 16.27 | 16.32 | -30.5% | -30.8% | -1.18 |
| Min_Above_020 | 2.50 | 3.14 | 3.65 | -20.4% | -31.5% | -0.70 |
| Metric | Tonight | 7d avg | 30d avg | Δ vs 7d | Δ vs 30d |
|---|---|---|---|---|---|
| Time in bed (min) | 390.9 | 386.3 | 366.8 | +1.2% | +6.6% |
| Total sleep (min) | 380.4 | 372.3 | 356.9 | +2.2% | +6.6% |
| Awake (min) | 10.5 | 14.0 | 9.9 | -24.9% | +6.1% |
| Deep (min) | 36.6 | 54.9 | 48.5 | -33.3% | -24.5% |
| REM (min) | 96.8 | 88.9 | 85.2 | +9.0% | +13.6% |
| Core (min) | 246.9 | 228.5 | 223.1 | +8.1% | +10.7% |
| Deep % | 9.63 | 14.79 | 14.42 | -34.9% | -33.2% |
| REM % | 25.46 | 23.71 | 24.61 | +7.4% | +3.5% |
| Core % | 64.91 | 61.50 | 60.98 | +5.5% | +6.5% |
| Sleep efficiency | 97.30 | 96.32 | 97.39 | +1.0% | -0.1% |
| Metric | Recent 7d | Prior 7d | Δ | Δ % |
|---|---|---|---|---|
| AHI | 1.550 | 2.411 | -0.861 | -35.7% |
| OA | 7.714 | 8.857 | -1.143 | -12.9% |
| H | 2.286 | 5.429 | -3.143 | -57.9% |
| CA | 0.143 | 0.429 | -0.286 | -66.7% |
| Pressure_Avg | 7.767 | 7.757 | +0.010 | +0.1% |
| Pressure_90th | 10.286 | 10.257 | +0.029 | +0.3% |
| Leak_Avg | 1.184 | 1.254 | -0.070 | -5.6% |
| Leak_90 | 4.629 | 3.943 | +0.686 | +17.4% |
| Leak_Max | 22.800 | 24.343 | -1.543 | -6.3% |
| Usage hours | 6.544 | 6.113 | +0.431 | +7.1% |
| FL_P95 | 0.091 | 0.103 | -0.011 | -11.1% |
| Min_Above_010 | 16.271 | 18.143 | -1.871 | -10.3% |
| Min_Above_020 | 3.143 | 3.557 | -0.414 | -11.6% |
| Date | AHI | OA | H | CA | PAvg | P90 | LAvg | LMax | L90 | FL_P95 | MinAbv010 | MinAbv020 | Hours |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026-09-24 | 1.65 | 3 | 6 | 1 | 7.36 | 10.18 | 0.59 | 28.80 | 1.20 | 0.09 | 15.2 | 3.4 | 6.05 |
| 2026-09-25 | 2.74 | 11 | 4 | 1 | 8.04 | 10.60 | 1.13 | 22.80 | 3.60 | 0.11 | 18.0 | 3.1 | 5.83 |
| 2026-09-26 | 2.72 | 13 | 6 | 0 | 7.66 | 10.50 | 2.37 | 22.80 | 9.60 | 0.10 | 19.2 | 5.1 | 6.98 |
| 2026-09-27 | 1.41 | 6 | 3 | 0 | 7.74 | 10.08 | 1.05 | 18.00 | 4.80 | 0.10 | 18.2 | 3.9 | 6.38 |
| 2026-09-28 | 1.23 | 7 | 1 | 0 | 7.54 | 10.36 | 1.04 | 34.80 | 3.60 | 0.11 | 22.2 | 3.5 | 6.48 |
| 2026-09-29 | 1.45 | 8 | 1 | 0 | 8.10 | 11.34 | 0.44 | 33.60 | 0.00 | 0.07 | 10.6 | 2.5 | 6.22 |
| 2026-09-30 | 1.44 | 7 | 2 | 0 | 8.01 | 10.40 | 1.18 | 16.80 | 3.60 | 0.10 | 16.9 | 2.0 | 6.25 |
| 2026-10-01 | 0.56 | 4 | 0 | 0 | 6.82 | 9.00 | 0.53 | 8.40 | 2.40 | 0.07 | 10.9 | 2.2 | 7.12 |
| 2026-10-02 | 2.04 | 9 | 3 | 1 | 8.50 | 10.32 | 1.68 | 25.20 | 8.40 | 0.09 | 15.9 | 2.8 | 6.38 |
| 2026-10-03 | 2.33 | 7 | 6 | 1 | 7.21 | 9.30 | 2.77 | 22.80 | 10.80 | 0.08 | 11.3 | 2.5 | 6.00 |
Based on 77 matched (CPAP + Apple Watch sleep-stage) nights through 2026-10-03 (76 prior + tonight). Correlations below are computed on the 31-night window (30d + tonight) that was fully re-staged for this run; the longer-horizon correlations from prior reports continue to apply.
| Pair | the user r (30d+tonight, n=31) | Interpretation |
|---|---|---|
| AHI vs deep_pct | +0.025 | essentially zero |
| AHI vs awake_min | -0.193 | weak negative |
| OA vs awake_min | -0.058 | negligible |
| Leak_Avg vs deep_pct | -0.055 | negligible |
| Leak_90 vs deep_pct | -0.056 | negligible |
| Leak_Max vs awake_min | -0.140 | weak negative |
| FL_P95 vs awake_min | -0.205 | weak negative |
| Min_Above_010 vs deep_pct | -0.041 | negligible |
Leak_90 median split (30d+tonight, median = 2.40): - Low-leak nights (n=18): mean deep% 14.64, awake 9.53 min - High-leak nights (n=13): mean deep% 13.74, awake 10.54 min
The pattern remains the same as prior reports: mild-to-moderate leak elevations show no demonstrable effect on sleep architecture in the aggregate. Tonight is a specific exception — leak was elevated (L90 = 10.80) AND deep% was unusually low (9.63%), which is directionally consistent with the weakly-negative Leak_90 vs deep_pct correlation but much larger in magnitude than the average effect. One night doesn't change the statistical story, but it's a sharper leak-plus-deep-loss pairing than the historical correlation implies.
Tonight is the follow-through on yesterday's leak warning, and this time the sleep architecture didn't save it.
The core story: Leak_90 and Leak_Avg are both >2.5 SD outliers vs the 30-day mean for the second straight night, with tonight actually worse than last night (L90 10.80 vs 8.40, LAvg 2.77 vs 1.68). Yesterday's report flagged this exact scenario — "if tomorrow's leak also returns to the 8–12 range, that's worth paying attention to" — and tomorrow has delivered it. This is now a credible mask-seal concern. Last night was a one-point outlier; tonight is pattern formation. The 3+ consecutive night threshold hasn't been crossed yet, but the next-night read will decide whether this is an incipient streak or a two-night blip.
Importantly, pressure did not rise to compensate tonight. Pressure_Avg 7.21 is the lowest of the last 10 nights, and Pressure_90th 9.30 is also the lowest (down from 10.32 last night). This differs from the usual auto-titrator response pattern and may be contributing to the hypopnea bump (H = 6, 7d z-equivalent > +1.4) and the lower deep-sleep yield. Not something the user can control directly — the machine's titration decides — but it's a mechanical note.
Sleep architecture reversed from last night's standout to tonight's weak-deep profile. Deep sleep crashed from 18.86% / 67.7 min last night to 9.63% / 36.6 min tonight — the lowest deep% in the recent 10-night window. REM held up at 25.46% (slightly above 30d average) and awake was fine (10.5 min, close to the 30d mean), and efficiency stayed high (97.30%). So the night wasn't fragmented — it was structurally shallow. This is directionally consistent with the known weakly-negative correlation between leak and deep sleep, but the magnitude is well above the average effect size.
The flagged metrics are specific: - Leak (2+ SD, two nights running) — this is the actionable signal. - Hypopnea count 6 (+1.40 SD) — more hypopneas, consistent with lower pressure and higher leak. - Deep sleep 9.63% — the lowest in the 10-night window, following a standout 18.86% last night.
What's not concerning: AHI 2.33 is still well below the clinical 5.0 threshold. REM is healthy. Sleep efficiency is 97.30%. The 7-day-over-7-day trend is strongly positive (-35.7% AHI). Usage was 6.0 h — on the low side vs 7d (6.54 h) but still compliant.
Action if there's one to take: next time the mask comes off the face, give the cushion/seal/headgear a quick inspection. If tomorrow's leak also lands in the 8–12 range, that would make 3 consecutive elevated-leak nights and would formally trigger the "3+ consecutive night" flag defined in this report series — at that point, mask replacement or re-fitting becomes a specific recommendation.