Sources: health-report-2026-08-03-to-2026-08-04.json (Apple Watch sleep stages, exported 06:37 on 08-04) + OSCAR_Greg_by_Day_2026-08-03.csv (ResMed AirSense/AirFlow 11 via OSCAR). Matched by calendar date of sleep-session start (22:28 on 08-03).
Three-night rising AHI trend, now above both trailing averages. AHI climbed 0.58 (08-01) → 1.16 (08-02) → 2.24 (08-03), roughly doubling each night. Tonight's 2.24 sits above the trailing 7-day average (1.59) and the trailing 30-day average (2.05), though it's still inside the range seen regularly earlier in the dataset (all-time mean: 2.31). The rise is driven mainly by a hypopnea spike — H = 8, tied for the 3rd-highest single-night count of all 43 matched nights (behind only 13 on 06-30 and 10 on 06-29) — plus central apneas (CA = 3, the 2nd-highest count on record, behind 4 on 07-31). Worth watching for one or two more nights to see if this is a genuine trend or noise; it isn't yet a sustained pattern (CA specifically has bounced between 1 and 3–4 rather than climbing steadily).
Deep sleep remains choppy and generally below the 30-day baseline (16.0%). Tonight's 12.1% is below both the 7-day (13.9%) and 30-day (16.0%) averages, continuing a mixed stretch: of the last 5 nights, 3 have landed below the 30-day deep-sleep baseline (07-30: 11.3%, 08-01: 5.8%, 08-03: 12.1%) against 2 above (07-31: 17.9%, 08-02: 14.8%). Not a sharp outlier on its own, but part of a softer deep-sleep run worth tracking.
Awake time and efficiency remain excellent. Awake time (10.5 min) and efficiency (97.5%) are both comfortably better than the 7-day (21.3 min / ~88%) and 30-day (18.9 min) averages, continuing the strong run since 08-02's near-best night. This is the reassuring counterpoint to the AHI/deep-sleep notes above — sleep continuity itself hasn't degraded.
| Metric | Value |
|---|---|
| Bedtime (session start) | 22:28 |
| Time in bed | 7.02 h (421.5 min) |
| Total sleep duration | 6.85 h (411.0 min) |
| Sleep efficiency (TST/TIB) | 97.5% |
| Awake time | 10.5 min — well below 7-day (21.3) and 30-day (18.9) averages |
| Core sleep | 4.22 h (253.4 min) — 61.7% |
| Deep sleep | 0.83 h (49.7 min) — 12.1% — below both trailing averages (7d: 13.9%, 30d: 16.0%) |
| REM sleep | 1.80 h (107.9 min) — 26.3% — above 30-day average (24.3%) |
| CPAP usage | 7.15 h |
| AHI | 2.24 — above 7-day (1.59) and 30-day (2.05) averages |
| RDI | 2.52 |
| OA / H / CA / UA / RERA | 5 / 8 / 3 / 0 / 2 — H and CA both near all-time highs |
| Pressure avg | 6.73 cmH₂O |
| Pressure 90th %ile | 10.14 cmH₂O |
| Leak avg | 0.70 L/min — moderate, in line with recent norm |
| Leak 90th %ile | 3.6 L/min |
| Date | AHI | OA | H | CA | RERA | Leak avg | Deep % | Awake (min) | Efficiency |
|---|---|---|---|---|---|---|---|---|---|
| 07-27 | 1.60 | 8 | 2 | 0 | 0 | 1.19 | 18.0% | 5.5 | 98.5% |
| 07-28 | 2.01 | 8 | 5 | 0 | 2 | 0.77 | 17.8% | 11.5 | 97.2% |
| 07-29 | 1.62 | 7 | 2 | 1 | 2 | 1.52 | 17.8% | 4.0 | 98.9% |
| 07-30 | 1.98 | 13 | 0 | 1 | 0 | 0.08 | 11.3% | 27.6 | 93.3% |
| 07-31 | 1.51 | 7 | 1 | 4 | 0 | 1.03 | 17.9% | 24.1 | 94.9% |
| 08-01 | 0.58 | 2 | 2 | 1 | 0 | 0.40 | 5.8% | 66.7 | 86.8% |
| 08-02 | 1.16 | 3 | 3 | 1 | 4 | 0.72 | 14.8% | 4.5 | 98.7% |
| 08-03 | 2.24 | 5 | 8 | 3 | 2 | 0.70 | 12.1% | 10.5 | 97.5% |
Split at leak ≤ 0.30 vs > 0.30 L/min (08-03's leak of 0.70 falls into the higher-leak bucket):
the user r (leak avg vs. deep %): −0.26 (n=43). the user r (leak avg vs. awake min): +0.00. the user r (AHI vs. awake min): +0.01. the user r (AHI vs. deep %): +0.05.
Interpretation: Tonight again falls in the higher-leak bucket, and deep sleep (12.1%) landed below both bucket means — consistent with the mild leak↔deep% relationship (r = −0.26) that has held steady across the dataset. Awake time, however, was excellent despite the higher-leak classification, reinforcing that leak level alone doesn't reliably predict sleep continuity — only deep-sleep share shows even a modest link.
| Metric | Tonight (08-03) | 7-day avg (07-28–08-03) | 30-day avg (07-05–08-03) |
|---|---|---|---|
| AHI | 2.24 | 1.59 | 2.05 |
| Leak avg (L/min) | 0.70 | 0.75 | 0.62 |
| Pressure avg (cmH₂O) | 6.73 | 6.94 | 7.87 |
| Usage hours | 7.15 | 7.13 | 6.73 |
| Deep sleep % | 12.1% | 13.9% | 16.0% |
| REM sleep % | 26.3% | 23.9% | 24.3% |
| Awake time (min) | 10.5 | 21.3 | 18.9 |
AHI and deep sleep % are both on the unfavorable side of their trailing averages tonight, while awake time, efficiency, REM%, and usage hours are all favorable or in line. Pressure (6.73) continues a multi-week downward drift — weekly averages have fallen steadily from ~8.5–8.75 cmH₂O in late June/early July to ~6.9–7.0 cmH₂O over the last two weeks, likely reflecting auto-titration settling rather than a new issue.
| Week | AHI avg | Leak avg | n |
|---|---|---|---|
| 06-22 to 06-28 | 2.84 | 1.98 | 7 |
| 06-29 to 07-05 | 3.60 | 1.49 | 6 |
| 07-06 to 07-12 | 3.00 | 0.43 | 7 |
| 07-13 to 07-19 | 2.13 | 0.45 | 6 |
| 07-20 to 07-26 | 1.43 | 0.71 | 7 |
| 07-27 to 08-02 | 1.49 | 0.82 | 7 |
| 07-28 to 08-03 (complete, 7 of 7) | 1.59 | 0.75 | 7 |
Trailing-week AHI (1.59) ticked up slightly from the prior week (1.49), driven almost entirely by tonight's 2.24 — the other six nights in the window were all ≤2.01. Leak eased slightly week over week (0.82 → 0.75). Both remain well within the range seen across the full dataset.
Watch AHI for one or two more nights, but no action indicated yet. Tonight's 2.24 breaks a run of low-AHI nights, driven by an unusually high hypopnea count (8) and central-apnea count (3) — both near the top of the 43-night history but not unprecedented (both have been matched or exceeded before, most recently CA=4 just three nights ago on 07-31). A single night doesn't establish a trend; the next few nights will clarify whether this is noise or a real shift.
Deep sleep is running softer than the 30-day baseline more often than not lately, though awake time and efficiency — arguably the more direct markers of sleep quality — remain strong. This split (good continuity, softer deep-sleep share) is worth keeping an eye on but doesn't yet look actionable.
Pressure continues its gradual multi-week decline (8.5+ → 6.9 cmH₂O), consistent with auto-adjusting therapy settling rather than anything to flag.