| Metric | Value |
|---|---|
| CPAP usage hours | 6.78 |
| Session count | 1 |
| AHI | 1.18 |
| RDI | 1.77 |
| Obstructive apneas (OA) | 6 |
| Hypopneas (H) | 2 |
| Central apneas (CA) | 0 |
| Unclassified apneas (UA) | 0 |
| RERA | 4 |
| Pressure_Avg | 7.90 |
| Pressure_Min | 4.00 |
| Pressure_Max | 11.30 |
| Pressure_90th | 10.18 |
| Leak_Avg | 1.91 |
| Leak_Max | 28.80 |
| Leak_90th | 4.80 |
| FL_Avg | 0.0241 |
| FL_P95 | 0.13 |
| FL_Max | 0.55 |
| Min_Above_005 (min) | 63.9 |
| Min_Above_010 (min) | 25.7 |
| Min_Above_020 (min) | 7.1 |
| Frac_NonZero (FL samples) | 35.53% |
| Apple Watch time in bed | 402.92 min |
| Apple Watch total sleep | 401.41 min |
| Apple Watch awake | 1.51 min |
| Apple Watch deep sleep | 31.61 min (7.88%) |
| Apple Watch REM sleep | 111.89 min (27.87%) |
| Apple Watch core sleep | 257.91 min (64.25%) |
| Sleep efficiency | 99.63% |
| Sleep session start | 2026-10-04T23:42:16-05:00 |
Trailing windows exclude tonight by construction (they cover the 7 and 30 nights ending 2026-10-03).
| Metric | Tonight | 7d avg | 30d avg | Ξ vs 7d | Ξ vs 30d | 30d z |
|---|---|---|---|---|---|---|
| AHI | 1.18 | 1.49 | 2.09 | -21.0% | -43.5% | -1.33 |
| RDI | 1.77 | 1.63 | 2.20 | +8.4% | -19.5% | -0.63 |
| OA count | 6.00 | 6.86 | 9.37 | -12.5% | -35.9% | -0.99 |
| H count | 2.00 | 2.29 | 3.13 | -12.5% | -36.2% | -0.53 |
| CA count | 0.00 | 0.29 | 0.73 | -100.0% | -100.0% | -0.76 |
| RERA count | 4.00 | 0.86 | 0.70 | +366.7% | +471.4% | +3.67 |
| Pressure_Avg | 7.90 | 7.70 | 7.51 | +2.6% | +5.3% | +0.91 |
| Pressure_90th | 10.18 | 10.11 | 10.03 | +0.6% | +1.5% | +0.29 |
| Leak_Avg | 1.91 | 1.24 | 0.94 | +53.9% | +102.9% | +1.36 |
| Leak_Max | 28.80 | 22.80 | 20.04 | +26.3% | +43.7% | +1.10 |
| Leak_90th | 4.80 | 4.80 | 3.04 | +0.0% | +57.9% | +0.53 |
| Usage hours | 6.78 | 6.40 | 6.37 | +5.9% | +6.5% | +1.08 |
| FL_Avg | 0.0241 | 0.0150 | 0.0150 | +61.0% | +60.3% | +2.98 |
| FL_P95 | 0.13 | 0.09 | 0.09 | +46.8% | +40.3% | +2.11 |
| FL_Max | 0.55 | 0.50 | 0.47 | +10.6% | +18.3% | +0.79 |
| Min_Above_005 | 63.90 | 37.53 | 36.87 | +70.3% | +73.3% | +3.19 |
| Min_Above_010 | 25.70 | 15.14 | 16.13 | +69.7% | +59.3% | +2.24 |
| Min_Above_020 | 7.10 | 2.77 | 3.63 | +156.2% | +95.4% | +2.14 |
| Frac_NonZero | 0.3553 | 0.2553 | 0.2427 | +39.2% | +46.4% | +2.35 |
| Metric | Tonight | 7d avg | 30d avg | Ξ vs 7d | Ξ vs 30d |
|---|---|---|---|---|---|
| Time in bed (min) | 402.92 | 379.27 | 379.42 | +6.2% | +6.2% |
| Total sleep (min) | 401.41 | 364.46 | 368.96 | +10.1% | +8.8% |
| Awake (min) | 1.51 | 14.80 | 10.46 | -89.8% | -85.6% |
| Deep (min) | 31.61 | 53.78 | 45.80 | -41.2% | -31.0% |
| REM (min) | 111.89 | 87.23 | 87.77 | +28.3% | +27.5% |
| Core (min) | 257.91 | 223.46 | 235.39 | +15.4% | +9.6% |
| Deep % | 7.88 | 14.78 | 12.45 | -6.90 pp | -4.57 pp |
| REM % | 27.87 | 23.78 | 23.70 | +4.09 pp | +4.17 pp |
| Core % | 64.25 | 61.44 | 63.84 | +2.81 pp | +0.41 pp |
| Sleep efficiency | 99.63 | 96.06 | 97.25 | +3.57 pp | +2.38 pp |
Read: This was a long, continuous, highly efficient night β ~402 min in bed with only 1.5 min awake. But the restorative composition shifted hard away from deep sleep (-31% vs 30d) toward REM (+27%). The pattern lines up cleanly with the elevated upper-airway resistance signals (FL/RERA): the breathing effort was present but was absorbed by the auto-titrator's pressure response and by stage reallocation, not by awakenings.
| Metric | Recent 7d | Prior 7d | Ξ | Ξ % |
|---|---|---|---|---|
| AHI | 1.461 | 2.229 | -0.767 | -34.4% |
| OA | 6.857 | 8.429 | -1.571 | -18.6% |
| H | 2.143 | 5.000 | -2.857 | -57.1% |
| CA | 0.286 | 0.429 | -0.143 | -33.3% |
| RERA | 1.286 | 0.857 | +0.429 | +50.0% |
| Pressure_Avg | 7.726 | 7.699 | +0.027 | +0.4% |
| Pressure_90th | 10.129 | 10.314 | -0.186 | -1.8% |
| Leak_Avg | 1.364 | 1.420 | -0.056 | -3.9% |
| Leak_90 | 4.800 | 4.971 | -0.171 | -3.4% |
| Leak_Max | 24.343 | 24.171 | +0.171 | +0.7% |
| Hours | 6.461 | 6.221 | +0.240 | +3.9% |
| FL_Avg | 0.0161 | 0.0156 | +0.0005 | +3.1% |
| FL_P95 | 0.0929 | 0.0986 | -0.0057 | -5.8% |
| Min_Above_010 | 16.21 | 17.00 | -0.79 | -4.6% |
| Frac_NonZero | 0.2698 | 0.2464 | +0.0233 | +9.5% |
AHI and H trend remains strongly improving week-over-week. The RERA bump (+50%) is driven almost entirely by tonight's spike of 4; prior nights were 1, 0, 0, 2, 1, 0, 1 β tonight 4 pulls the average up single-handedly.
| Date | AHI | OA | H | CA | RERA | PAvg | P90 | LAvg | L90 | FL_P95 | MinAbv010 | FracNZ | Hours | Deep% |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026-09-25 | 2.74 | 11 | 4 | 1 | 1 | 8.04 | 10.60 | 1.13 | 3.60 | 0.11 | 18.0 | 0.251 | 5.83 | 13.14 |
| 2026-09-26 | 2.72 | 13 | 6 | 0 | 2 | 7.66 | 10.50 | 2.37 | 9.60 | 0.10 | 19.2 | 0.189 | 6.98 | 11.66 |
| 2026-09-27 | 1.41 | 6 | 3 | 0 | 0 | 7.74 | 10.08 | 1.05 | 4.80 | 0.10 | 18.2 | 0.254 | 6.38 | 12.85 |
| 2026-09-28 | 1.23 | 7 | 1 | 0 | 1 | 7.54 | 10.36 | 1.04 | 3.60 | 0.11 | 22.2 | 0.237 | 6.48 | 13.79 |
| 2026-09-29 | 1.45 | 8 | 1 | 0 | 0 | 8.10 | 11.34 | 0.44 | 0.00 | 0.07 | 10.6 | 0.220 | 6.22 | 13.30 |
| 2026-09-30 | 1.44 | 7 | 2 | 0 | 0 | 8.01 | 10.40 | 1.18 | 3.60 | 0.10 | 16.9 | 0.326 | 6.25 | 16.10 |
| 2026-10-01 | 0.56 | 4 | 0 | 0 | 2 | 6.82 | 9.00 | 0.53 | 2.40 | 0.07 | 10.9 | 0.235 | 7.12 | 16.96 |
| 2026-10-02 | 2.04 | 9 | 3 | 1 | 2 | 8.50 | 10.32 | 1.68 | 8.40 | 0.09 | 15.9 | 0.327 | 6.38 | 18.86 |
| 2026-10-03 | 2.33 | 7 | 6 | 1 | 2 | 7.21 | 9.30 | 2.77 | 10.80 | 0.08 | 11.3 | 0.188 | 6.00 | 9.63 |
| 2026-10-04 | 1.18 | 6 | 2 | 0 | 4 | 7.90 | 10.18 | 1.91 | 4.80 | 0.13 | 25.7 | 0.355 | 6.78 | 7.88 |
Based on 100 matched (CPAP + Apple Watch sleep-stage) nights through 2026-10-04. Correlations below are computed on the 30d matched window (CPAP dates 2026-09-05 to 2026-10-04, n=27), now including tonight. Tonight's night sits at the extreme of the FL distribution so it meaningfully influences the FLβdeep correlation; the pre-tonight 30d (n=26) estimates are retained in prior reports.
| Pair | the user r (30d, n=26) | Interpretation |
|---|---|---|
| AHI vs deep_pct | -0.026 | essentially zero |
| AHI vs awake_min | -0.305 | weak negative |
| OA vs awake_min | -0.111 | negligible |
| Leak_Avg vs deep_pct | +0.080 | negligible |
| Leak_90th vs deep_pct | +0.062 | negligible |
| Leak_Max vs awake_min | +0.111 | negligible |
| FL_P95 vs awake_min | -0.319 | weak negative |
| FL_P95 vs deep_pct | +0.179 | weak positive (noisy) |
| FL_Avg vs deep_pct | +0.391 | weak-moderate positive |
| Min_Above_010 vs deep_pct | +0.097 | negligible |
| RERA vs awake_min | -0.119 | negligible |
| RERA vs deep_pct | -0.096 | negligible |
Leak_90th median split (30d matched, median = 2.40): - Low-leak nights (n=16): mean deep% = 12.32, mean awake = 10.00 min - High-leak nights (n=10): mean deep% = 12.84, mean awake = 10.77 min
Essentially the same pattern as prior reports: leak elevations within the user's observed range don't meaningfully move sleep architecture on average. The 30d correlations continue to point to very small effect sizes across the board. The one semi-interesting line is FL_Avg vs deep_pct at r = +0.391, which is unexpected directionally (more flow limitation β more deep sleep) and is likely noise given n=26; it will be worth checking once tonight's elevated-FL night gets its stage data pair.
Tonight is a quiet night by AHI measures but a loud night by flow-limitation and RERA measures. The frank-event picture (AHI, OA, H, CA) is the best in the last 10 nights except for 2026-10-01: AHI 1.18 is well below the 30d mean, there were no central apneas, and only 2 hypopneas. By any clinical threshold this is a good night.
But the airway-resistance picture β the metrics that measure subthreshold breathing effort β is the worst of the recent 31-night window. Every single flow-limitation metric is in the top-3 of the window: FL_Avg rank 1 (z = +2.98), Min_Above_005 rank 1 (z = +3.19), Frac_NonZero rank 1 (z = +2.35). The RERA count of 4 is similarly rank 1 (z = +3.67). Both measures capture respiratory effort that the airway partially resists but doesn't fully occlude β the kind of pattern that is often associated with position changes (back vs side), nasal congestion, or a slightly different sleep stage distribution. The auto-titrator responded appropriately β Pressure_Avg climbed to 7.90 (up from 7.21 last night) and Pressure_90th to 10.18 (up from 9.30) β which likely kept the AHI low. Said differently: the machine saw the resistance coming and pushed harder, successfully preventing it from becoming apneas, but the resistance itself still shows up in the FL/RERA metrics.
The leak concern from last night is cleared. Yesterday's report explicitly watched for a 3rd consecutive elevated-leak night as the trigger for a mask-fit recommendation. Tonight's Leak_90 came back to 4.80 (within normal), and the 3-night streak did not form. Leak_Avg is still modestly elevated at 1.91 (+1.36 SD), but the pattern has broken β no action needed on mask seal.
Sleep architecture now in hand β the FL/RERA signal did cost deep sleep. The Apple Watch session for the night of 2026-10-04 (session start 2026-10-04T23:42:16) is now exported. Deep sleep came in at 31.6 min / 7.88%, which is -31.0% vs 30d avg (45.8 min / 12.45%) and -41.2% vs the recent 7d avg (53.8 min / 14.78%). This is the second-lowest Deep% in the recent window β only 2026-10-03 was lower (9.63%). So the question "did the FL/RERA spike cost the user deep sleep?" answers cleanly: yes, notably. Deep sleep was reallocated to REM (+27β28% vs averages), and sleep continuity was in fact better than usual (awake only 1.51 min, efficiency 99.63%). The auto-titrator responded to the airway resistance well enough to prevent arousals and awakenings, but not well enough to preserve slow-wave sleep architecture.
Is this actionable? On its own, no β one night of reduced deep sleep after a lifetime of normal deep sleep isn't a problem. But the pairing is clinically interesting: elevated flow-limitation and RERA signals correlated with a ~31% reduction in deep-sleep minutes on the same night, with no corresponding reduction in overall sleep duration or efficiency. That's exactly the shape of thing that would be invisible if we only tracked AHI. Watch next 2β3 nights: if FL/RERA stays elevated AND deep sleep stays suppressed, that's a pattern worth investigating (nasal congestion? seasonal allergies? different sleep position? mask cushion aging in a way that increases resistance without yet causing leak?). If tomorrow's night shows FL/RERA back to baseline AND deep sleep recovers, this was a one-off and the correlation is a useful calibration data point for future monitoring. The new matched night (n=27 on 30d matched window) is the first data point actually at the upper tail of the FL distribution β all prior FL correlation estimates were effectively interpolation; tonight tests whether the signal is real at extremes.
health-report-2026-10-04-to-2026-10-05.json, this report regenerated after original issuance). Note the session start time (2026-10-04T23:42:16-05:00) is a few minutes before midnight β slightly earlier than the usual pairing convention of "CPAP date X pairs with the Apple Watch sleep session starting after midnight on X+1". The CPAP usage hours (6.78h) and Apple Watch time-in-bed (6.72h) line up to within 4 minutes, so the pairing is sound. This earlier bedtime is consistent with the long TIB tonight.