Sources: health-report-2026-08-02-to-2026-08-03.json (Apple Watch sleep stages, exported 08:36 on 08-03) + OSCAR_Greg_by_Day_2026-08-02.csv (ResMed AirSense/AirFlow 11 via OSCAR). Matched by calendar date of sleep-session start (23:02 on 08-02).
Sharpest single-night reversal in the dataset. Last night (08-01) was the most disrupted night on record — 66.7 min awake, 86.8% efficiency, 5.8% deep sleep. Tonight flipped almost completely: awake time fell to 4.5 min (2nd-lowest of 42 matched nights, behind only 4.0 min on 07-29) and efficiency hit 98.7% (3rd-highest on record). Deep sleep rebounded to 14.8%, back near the 30-day baseline (16.5%). This argues the 08-01 spike was a one-off rather than the start of a sustained trend — worth confirming with another night or two, but the immediate reversal is a good sign.
RERA count (4) ties the dataset's highest recorded value (previously seen on 07-03 and 07-15), roughly 4x the all-time mean of ~1.0. This pushed RDI (1.83) well above AHI (1.16) tonight. Notably, this did not coincide with disrupted sleep — awake time and efficiency were both near-best. Consistent with the broader pattern in this data: individual event-type counts don't reliably predict how the night feels architecturally.
The 3-night rising-awake-time trend flagged after 08-01 (27.6 → 24.1 → 66.7 min) is broken — tonight's 4.5 min is a sharp drop, not a continuation.
| Metric | Value |
|---|---|
| Bedtime (session start) | 23:02 |
| Time in bed | 5.90 h (354.2 min) |
| Total sleep duration | 5.83 h (349.7 min) |
| Sleep efficiency (TST/TIB) | 98.7% — 3rd-highest in the dataset |
| Awake time | 4.5 min — 2nd-lowest in the dataset (lowest: 4.0 min on 07-29) |
| Core sleep | 3.37 h (202.2 min) — 57.8% |
| Deep sleep | 0.86 h (51.7 min) — 14.8% — near 30-day average (16.5%) |
| REM sleep | 1.60 h (95.8 min) — 27.4% — above 30-day average (24.5%) |
| CPAP usage | 6.02 h |
| AHI | 1.16 — below 30-day average (2.06) |
| RDI | 1.83 |
| OA / H / CA / UA / RERA | 3 / 3 / 1 / 0 / 4 — RERA ties all-time high |
| Pressure avg | 7.8 cmH₂O |
| Pressure 90th %ile | 10.3 cmH₂O |
| Leak avg | 0.72 L/min — moderate |
| Leak 90th %ile | 1.2 L/min |
| Date | AHI | OA | RERA | Leak avg | Leak 90th/95th | Deep % | Awake (min) | Efficiency |
|---|---|---|---|---|---|---|---|---|
| 07-26 | 1.24 | 5 | 0 | 0.39 | 0.0 | 14.3% | 26.6 | 92.9% |
| 07-27 | 1.60 | 8 | 0 | 1.19 | 2.4 | 18.0% | 5.5 | 98.5% |
| 07-28 | 2.01 | 8 | 2 | 0.77 | 2.4 | 17.8% | 11.5 | 97.2% |
| 07-29 | 1.62 | 7 | 2 | 1.52 | 6.0 | 17.8% | 4.0 | 98.9% |
| 07-30 | 1.98 | 13 | 0 | 0.08 | 0.0 | 11.3% | 27.6 | 93.3% |
| 07-31 | 1.51 | 7 | 0 | 1.03 | 3.6 | 17.9% | 24.1 | 94.9% |
| 08-01 | 0.58 | 2 | 0 | 0.40 | 2.4 | 5.8% | 66.7 | 86.8% |
| 08-02 | 1.16 | 3 | 4 | 0.72 | 1.2 | 14.8% | 4.5 | 98.7% |
Split at leak ≤ 0.30 vs > 0.30 L/min (08-02's leak of 0.72 falls into the higher-leak bucket):
the user r (leak avg vs. deep %): −0.27 (n=42). the user r (leak avg vs. awake min): 0.00. the user r (AHI vs. awake min): 0.00. the user r (AHI vs. deep %): +0.04.
Interpretation: Tonight sits in the higher-leak bucket by a hair (0.72 vs. the 0.30 cutoff) yet posted one of the best awake/efficiency results in the whole dataset — another data point reinforcing that leak level alone doesn't predict a disrupted night. The leak↔deep% relationship remains the one modestly consistent signal (r = −0.27); leak↔awake and AHI↔awake remain essentially flat across 42 nights.
| Metric | Tonight (08-02) | 7-day avg (07-27–08-02) | 30-day avg (07-04–08-02) |
|---|---|---|---|
| AHI | 1.16 | 1.49 | 2.06 |
| Leak avg (L/min) | 0.72 | 0.82 | 0.59 |
| Pressure avg (cmH₂O) | 7.8 | 6.96 | 7.95 |
| Usage hours | 6.02 | 7.00 | 6.68 |
| Deep sleep % | 14.8% | 14.8% | 16.5% |
| REM sleep % | 27.4% | 23.2% | 24.5% |
| Awake time (min) | 4.5 | 20.6 | 17.8 |
Tonight's usage (6.02 h) is the shortest session of the trailing week — despite that, sleep quality metrics (awake time, efficiency) were among the best in the dataset, underscoring again that shorter, higher-quality sleep opportunities aren't necessarily worse than longer ones (see 08-01's inverse case). AHI and leak are both inside normal range; pressure (7.8) is above the 7-day average but below the 30-day average.
| Week | AHI avg | Leak avg | n |
|---|---|---|---|
| 06-22 to 06-28 | 2.84 | 1.98 | 7 |
| 06-29 to 07-05 | 3.24 | 1.41 | 7 |
| 07-06 to 07-12 | 3.00 | 0.43 | 7 |
| 07-13 to 07-19 | 2.22 | 0.42 | 7 |
| 07-20 to 07-26 | 1.43 | 0.71 | 7 |
| 07-27 to 08-02 (complete, 7 of 7) | 1.49 | 0.82 | 7 |
Trailing-week AHI (1.49) is essentially flat vs. the prior week (1.43) and remains near the lowest levels seen since tracking began. Leak ticked up slightly week over week (0.71 → 0.82) but stays well within the range seen across the dataset.
The 08-01 anomaly looks like a one-off, not a new trend. Tonight's near-total reversal — from the worst night on record to a top-3 night on efficiency and a top-2 night on awake time — happened within 24 hours and without any corresponding change in CPAP performance (AHI and leak were unremarkable both nights). This is the strongest evidence yet that whatever drove 08-01's disruption (most likely tied to that night's unusually long time-in-bed) was situational rather than a shift in underlying sleep or therapy quality.
RERA count (4) is worth a light watch, not a flag. It ties the dataset's high, but two prior instances of RERA=4 (07-03, 07-15) didn't correspond to notably bad nights either, so this isn't yet a pattern with a downstream effect.
AHI, leak, and pressure remain in a healthy, stable range heading into the next week — no action indicated on the therapy side.