Sources: health-report-2026-08-21-to-2026-08-22.json (Apple Watch sleep stages, exported 06:09 on 08-22) + OSCAR_Greg_by_Day_2026-08-21.csv (ResMed AirSense/AirFlow 11 via OSCAR). Session start was 22:17 on 08-21 — 50 minutes earlier than 08-20's 23:07, reverting to a pre-cluster-worst bedtime and ending 08-20's late-bedtime, short-TIB night. CPAP hours 7.80 vs. Watch TIB 7.65 (Δ ≈ 8.9 min, CPAP longer — machine ran briefly past Watch session close; tight alignment).
AHI CLUSTER HARDENING — 4TH CONSECUTIVE RISING NIGHT. ESCALATE TO SLEEP CLINICIAN NOW. 4-night AHI trajectory: 1.63 → 2.20 → 2.55 → 2.82 (monotonic, +73% over four nights). 4-night OA trajectory: 7 → 10 → 14 → 15 (monotonic, +114%). Tonight's AHI 2.82 is rank 32/34 of matched history — 2nd-highest AHI in the matched window. Tonight's OA count 15 is rank 33/34 — the 2nd-highest OA count in matched history. OA/hr = 1.92, vs. 30-day baseline of 1.15/h (+67%). The 08-20 report set an explicit escalation threshold: "if 08-19/20/21 trajectory continues (2.20 → 2.55 → higher), this is a hardening cluster and warrants clinical escalation this week rather than 'on the next check-in.'" That threshold is now crossed. The cluster is no longer a 2–3 night anomaly; it is a monotonic 4-night trend spanning both leaky and leak-fixed nights. Note also H = 6 tonight (rank 32/34) — tonight is the highest hypopnea count of the current cluster window, and combined with 15 OA, the total obstructive event load (OA + H = 21) is the highest of the current window.
MASK FIX HELD AND IMPROVED — MECHANICAL ISSUE FULLY RESOLVED. Leak_avg dropped to 0.20 L/min (2nd consecutive very-low night: 08-20 was 0.69). This is rank 5/34 of matched history — 5th-lowest leak_avg ever recorded. Leak_90 = 0.0 (rank 1/34, tied at the floor). Leak_max = 10.8, down from 15.6 on 08-20 and 26.4 on 08-19. The 08-20 report's exit criterion — "if leak stays ≤ 0.5 for 3 consecutive nights, the mechanical issue can be considered fully resolved" — is at 2-of-3 (0.69, 0.20), and both nights are trending well below the ≥1.0 mechanical-issue threshold. The mask intervention is a decisive win and can be considered essentially closed pending 08-22 confirmation.
THE DECOUPLING HYPOTHESIS IS CONFIRMED — TONIGHT IS THE CLEANEST TEST YET. The worst obstructive-event-load night in matched history (excluding 06-26 pre-cluster outlier) co-occurred with the best-quality-sleep night in matched history: TST 7.61 h, awake time 2.51 min (rank 1/34 — the lowest awake time ever recorded), efficiency 99.45% (rank 34/34 — the highest efficiency ever recorded), Deep 14.18% (at the 30-day baseline of 14.19% almost exactly), REM 25.49% (above 30-day 23.91%). The 08-20 report asked for exactly this test: "If Deep ≥12% on 08-21, tonight [08-20's 7.68%] was noise. If Deep <10% again, the coupling story becomes plausible..." Deep landed at 14.18%, well above the 12% threshold. 08-20's deep-sleep collapse was noise, not the start of an event-load-to-architecture coupling. The architecture channel and the CPAP-mechanics channel remain independently governed — the elevated obstructive load is not manifesting in Watch-visible sleep-quality degradation.
PRESSURE PARTIALLY RECOVERED DESPITE HIGHER EVENT LOAD. P_avg pulled back slightly from 7.80 → 7.68 (near 30-day 7.41), and P_90 dropped 11.16 → 10.14 (rank 17/34, right at the median; near 30-day 10.03). This is odd given tonight's event count is higher than 08-20's: normally we'd expect the auto-titrator to run harder on a higher-OA night. The plausible reading: tonight's 15 OA + 6 H events were more discrete and shorter (not a sustained obstructive demand pattern), so the algorithm didn't need sustained pressure elevation to work through them. This is not consistent with a worsening underlying airway obstruction; it is more consistent with intermittent event bursts. Something to raise with the clinician alongside the event-count data.
LONGEST USE + LONGEST TIB + BEST QUALITY IN 11+ NIGHTS. CPAP hours 7.80 = rank 32/34 (near-longest ever). TIB 7.65 h = rank 32/34 (also near-longest). Bedtime moved 50 min earlier than 08-20 (22:17 vs. 23:07). This is the first night in the current cycle where TIB, TST, efficiency, and awake time are all simultaneously in best-of-history territory. The sleep experience itself was, by every architecture metric, the best night of the matched window.
| Metric | Value |
|---|---|
| Bedtime (session start) | 22:17 on 08-21 (−50 min vs. 08-20's 23:07) |
| Time in bed | 7.65 h (459.1 min) — rank 32/34; +59.5 min vs. 30-day (399.6) |
| Total sleep duration | 7.61 h (456.6 min) — best of matched history at this position |
| Sleep efficiency (TST/TIB) | 99.45% — rank 34/34 (highest ever recorded); +4.02 pt vs. 30-day (95.44%) |
| Awake time | 2.51 min — rank 1/34 (lowest ever recorded); −87% vs. 30-day (18.80) |
| Core sleep | 4.59 h (275.5 min) — 60.33% (at 30-day norm of 61.90%) |
| Deep sleep | 1.08 h (64.7 min) — 14.18% — recovered from 08-20's 7.68%; essentially = 30-day baseline (14.19%) |
| REM sleep | 1.94 h (116.4 min) — 25.49% — +4.58 pts vs. 08-20 (20.91%); +1.58 pts vs. 30-day (23.91%) |
| CPAP usage | 7.80 h — rank 32/34; +24.4% vs. 08-20 (6.27); +15.8% vs. 30-day (6.74) |
| AHI | 2.82 — rank 32/34 (2nd-highest of matched history); +11% vs. 08-20 (2.55); +65% vs. 30-day (1.70) |
| RDI | 2.82 (RERA = 0) |
| OA / H / CA / UA / RERA | 15 / 6 / 1 / 0 / 0 — OA rank 33/34 (2nd-highest ever); H rank 32/34 (2nd-highest ever) |
| OA per hour | 1.92 /h — +67% vs. 30-day baseline (1.15/h); near the highest OA/h rates of matched history |
| Pressure avg | 7.68 cmH₂O — −0.12 vs. 08-20 (7.80); +0.27 vs. 30-day (7.41); rank 21/34 |
| Pressure 90th %ile | 10.14 cmH₂O — −1.02 vs. 08-20 (11.16); +0.11 vs. 30-day (10.03); rank 17/34 (right at median) |
| Pressure max | 11.12 cmH₂O |
| Leak avg | 0.20 L/min — −71% vs. 08-20 (0.69); rank 5/34 (5th-lowest of matched history) |
| Leak 90th %ile | 0.0 L/min — rank 1/34 (tied at floor); leak channel at absolute floor |
| Leak max | 10.8 L/min — −31% vs. 08-20 (15.6) |
| Compliance flag | 1 (compliant) |
| Date | AHI | OA | H | CA | RERA | Leak avg | Leak 90 | P_avg | P_90 | Hours | Deep % | REM % | Awake (min) | Eff |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 08-11 | 2.57 | 12 | 4 | 0 | 0 | 0.42 | 1.2 | 7.28 | 10.02 | 6.62 | 13.57 | 25.24 | 21.1 | 94.61% |
| 08-12 | 2.68 | 13 | 3 | 2 | 1 | 0.28 | 1.2 | 7.99 | 10.00 | 6.72 | 20.78 | 19.11 | 28.1 | 92.80% |
| 08-13 | 1.29 | 8 | 0 | 2 | 0 | 0.20 | 1.2 | 7.01 | 9.54 | 7.75 | 10.93 | 26.70 | 20.1 | 95.60% |
| 08-14 | 1.63 | 11 | 1 | 0 | 1 | 0.22 | 0.0 | 7.94 | 10.86 | 7.38 | 9.73 | 29.20 | 33.6 | 92.19% |
| 08-16 | 1.13 | 3 | 3 | 1 | 2 | 0.91 | 4.8 | 7.92 | 10.94 | 6.22 | 14.73 | 19.78 | 6.0 | 98.34% |
| 08-17 | 1.11 | 5 | 2 | 0 | 0 | 1.14 | 1.2 | 7.57 | 9.54 | 6.28 | 14.21 | 25.35 | 7.5 | 97.95% |
| 08-18 | 1.63 | 7 | 2 | 1 | 1 | 1.19 | 3.6 | 7.55 | 11.11 | 6.12 | 8.86 | 18.71 | 13.5 | 96.34% |
| 08-19 | 2.20 | 10 | 5 | 0 | 0 | 1.92 | 4.8 | 8.18 | 10.34 | 6.83 | 16.50 | 24.11 | 7.0 | 98.25% |
| 08-20 | 2.55 | 14 | 2 | 0 | 0 | 0.69 | 1.2 | 7.80 | 11.16 | 6.27 | 7.68 | 20.91 | 13.0 | 96.43% |
| 08-21 | 2.82 | 15 | 6 | 1 | 0 | 0.20 | 0.0 | 7.68 | 10.14 | 7.80 | 14.18 | 25.49 | 2.51 | 99.45% |
(08-15 not in table: no matched Watch sleep-stage session — CPAP-only that night was AHI 1.53, OA 8, Leak 0.31, P_avg 8.35, Hours 5.88.)
3-night composite (08-19 → 08-21): AHI mean 2.52 (+48% vs. 30-day 1.70) • OA total 39 across 20.90 h use = 1.87 OA/h (+62% vs. 30-day 1.15/h) • Deep mean 12.79% (−1.40 pts vs. 30-day 14.19% — modestly lower but within noise) • REM mean 23.51% (essentially at 30-day 23.91%) • Awake mean 7.5 min (much better than 30-day 18.80). Event-load channel is severely elevated; architecture channel is at baseline or better. This is the cleanest instance in the matched window of the two channels moving independently in opposite directions.
Leak trajectory closure (7-night matched, 08-16 → 08-21): 0.91 → 1.14 → 1.19 → 1.92 → 0.69 → 0.20. Two consecutive nights well under 0.5 confirm the mask fix stuck. Leak_90 has dropped 4.8 → 1.2 → 0.0 across the last 3 nights — the tail is fully controlled. The mechanical/delivery channel is definitively fixed. The 08-20 report's 3-night confirmation criterion needs one more clean night (08-22) to close entirely, but on trajectory this is a done deal.
4-night AHI escalation trajectory (08-18 → 08-21):
| Date | AHI | OA | H | OA+H | OA/h | Leak_avg | Interpretation |
|---|---|---|---|---|---|---|---|
| 08-18 | 1.63 | 7 | 2 | 9 | 1.14 | 1.19 | Leaky, moderate events |
| 08-19 | 2.20 | 10 | 5 | 15 | 1.46 | 1.92 | Peak leak, rising events |
| 08-20 | 2.55 | 14 | 2 | 16 | 2.23 | 0.69 | Leak-fixed, events still rising |
| 08-21 | 2.82 | 15 | 6 | 21 | 1.92 | 0.20 | Leak eliminated, events still rising |
Interpretation. Event load has risen every night for four consecutive nights, across a range of leak states — from 1.92 down to 0.20. The event-load elevation is fully independent of leak. The 08-19 "leak fix natural experiment" now has a longer follow-through: the mask fix continued to work, and the AHI/OA cluster continued to escalate anyway. This is no longer a hypothesis-in-doubt; it is a firmly established finding across four nights and a >90% reduction in leak. Something else — physiology, position, congestion, weight, alcohol, medication, upper-airway inflammation, reflux, or natural airway-tone drift — is driving this.
| 08-20 report prediction for 08-21 | Actual 08-21 outcome | Verdict |
|---|---|---|
| AHI drops back below 2.0 = "08-20 was local peak" | AHI 2.82 (highest of cluster) | Prediction FAILED — cluster is hardening, not peaking |
| Leak stays ≤ 0.5 = "mask fix held" | Leak 0.20 | Prediction MET — fix held and improved |
| P_90 cools with AHI (coupled) OR stays high (algorithm sees more than counter) | P_90 dropped to 10.14 despite higher AHI | Uncoupled — algorithm did not respond to higher event count; suggests shorter/more discrete events |
| Deep ≥ 12% closes the coupling concern | Deep 14.18% | Prediction MET — deep back to baseline, coupling concern closed |
| Deep < 10% opens coupling concern further | n/a — deep at 14.18% | Coupling concern closed |
| Watch for shorter TIB continuing to compress deep-minutes | TIB 7.65 h (longest of last 11 nights) | Bedtime moved earlier; no TIB compression |
Net verdict: 4 of 5 predictions met or resolved. The one that failed (AHI drops back) is now the primary clinical concern — the cluster is not self-resolving.
Buckets on nightly leak average:
the user correlations (n=34, tonight included): leak_avg vs. deep_pct = −0.03 (essentially zero); leak_avg vs. awake_min = −0.28 (weak negative — high-leak nights tend to have less awake time, counter to intuition); ahi vs. deep_pct = −0.01 (essentially zero); p_90 vs. awake_min = −0.44 (moderate negative — higher pressure associates with less awake time); ahi vs. awake_min = −0.18 (weak negative — high-AHI nights have slightly less awake time, again counter to intuition). The full-dataset finding stands: no strong CPAP-mechanics-to-architecture linkage. Tonight's data point — worst events, best sleep — is the strongest single-night confirmation of that decoupling.
| Metric | Tonight (08-21) | 7-day avg (through 08-20) | 30-day avg (through 08-20) |
|---|---|---|---|
| AHI | 2.82 | 1.65 | 1.70 |
| OA count | 15 | 8.29 | 7.77 |
| OA per hour | 1.92 /h | ~1.24 /h | ~1.15 /h |
| H count | 6 | 2.14 | 2.73 |
| Leak avg (L/min) | 0.20 | 0.90 | 0.84 |
| Leak 90th (L/min) | 0.0 | 2.40 | 2.60 |
| Pressure avg (cmH₂O) | 7.68 | 7.71 | 7.41 |
| Pressure 90th (cmH₂O) | 10.14 | 10.50 | 10.03 |
| Usage hours | 7.80 | 6.69 | 6.74 |
| Deep sleep % | 14.18% | 11.81% | 14.19% |
| REM sleep % | 25.49% | 23.54% | 23.91% |
| Awake time (min) | 2.51 | 14.41 | 18.80 |
| Efficiency | 99.45% | 96.45% | 95.44% |
| TIB (h) | 7.65 | 6.56 | 6.66 |
Verdict: extreme divergence across channels. The event channel is at cluster-worst (AHI +65% vs. 30-day, OA/h +67% vs. 30-day). The delivery channel is at cluster-best (leak_avg −76% vs. 30-day, leak_90 at floor). The architecture channel is at all-time-best (efficiency and awake time both set new records; deep and REM at or above baseline). The pressure channel is at baseline (P_avg and P_90 both essentially at 30-day norms). This is the strongest possible confirmation that the AHI escalation is a real, independent, physiological phenomenon that is not (a) a mask-seal artifact, (b) associated with sleep-quality degradation, or (c) being fully compensated by the auto-titrator's pressure response.
| Week (Wed–Tue) | AHI avg | Leak avg | P_avg | P_90 avg | OA avg | Hours | n |
|---|---|---|---|---|---|---|---|
| 07-16 to 07-21 | 1.62 | 0.46 | 8.18 | 10.78 | 6.7 | 6.38 | 6 |
| 07-22 to 07-28 | 1.68 | 0.87 | 7.41 | 10.22 | 8.6 | 6.76 | 7 |
| 07-29 to 08-04 | 1.74 | 0.70 | 7.03 | 9.27 | 7.3 | 7.12 | 7 |
| 08-05 to 08-11 | 1.83 | 1.16 | 7.51 | 10.13 | 7.6 | 6.37 | 7 |
| 08-12 to 08-18 | 1.57 | 0.61 | 7.76 | 10.34 | 7.9 | 6.62 | 7 |
| 08-19 to 08-21 (partial, n=3) | 2.52 | 0.94 | 7.89 | 10.55 | 13.0 | 6.97 | 3 |
The partial current week (3 nights) now has AHI 2.52 — highest weekly average since the pressure-instrumented window began — and OA of 13.0/night, ~64% above the trailing 5-week peak of 8.6. This is on pace to be the worst week for obstructive event load by a wide margin. Even if 08-22 through 08-25 revert fully to 30-day baseline (AHI 1.70), the week's average will still be ~2.05, which would beat all 5 prior weeks.
PRIMARY FINDING: TIME TO CONTACT THE SLEEP CLINICIAN. The 08-20 report set a threshold — "if 08-19/20/21 trajectory continues, escalate this week." That threshold is met. Package for the clinician: 4-night AHI trajectory 1.63 → 2.20 → 2.55 → 2.82 (monotonic, +73%); OA trajectory 7 → 10 → 14 → 15 (monotonic, +114%); this occurred across a leak-fix intervention that took leak from 1.92 down to 0.20, so mechanical/mask causes are effectively ruled out; sleep architecture is at or above baseline (deep, REM, efficiency all normal-to-best), so this is not producing subjective sleep degradation and would not be flagged by symptom-based screening alone; auto-pressure is at baseline (P_avg 7.68, P_90 10.14), so the machine is not straining. The relevant question for the clinician is what could cause a sustained obstructive-event elevation in someone with fixed leak, baseline pressure, and normal sleep architecture. Candidate hypotheses (unchanged from 08-20 outlook): recent weight change, positional shift (side↔back), nasal congestion/seasonal allergy load, alcohol/sedative intake, new medication, upper-airway inflammation, reflux, or natural airway-tone drift. Also worth adding: (a) the timing coincides with late-summer allergen load in the Midwest; (b) the pattern started around 08-11, so anything changed in ~08-10 is a candidate.
MASK CHANNEL: ESSENTIALLY CLOSED. Leak_avg 0.20 tonight (best of last 11 nights, rank 5/34 lifetime). Leak_90 at 0.0. One more clean night on 08-22 finishes the 3-night confirmation criterion the 08-20 report specified. Continue with the current mask setup and stop watching this channel actively unless a leak rebound appears.
PRESSURE CHANNEL: BENIGN. P_90 dropped despite higher event counts. This is a mild positive: the auto-titrator is not straining. Watch P_90 for the next 2–3 nights — if it stays ≤ 10.5 while AHI is elevated, the algorithm is confirming that tonight's obstructive events were the discrete/short-burst variety rather than sustained pressure-demanding obstruction. If P_90 climbs back to ≥ 11 while AHI is also elevated, that would upgrade the concern (sustained obstruction rather than intermittent).
ARCHITECTURE CHANNEL: DECOUPLING CONFIRMED. Tonight's best-ever efficiency and lowest-ever awake time despite worst-of-cluster events settles the coupling question in favor of independence. Stop worrying about deep-sleep sensitivity to the AHI cluster. The 08-20 low deep read was noise. This also means the AHI cluster is unlikely to be urgent from a sleep-quality standpoint — it is a quiet clinical elevation that only shows up in event counts, which is precisely the kind of finding CPAP monitoring is designed to catch that symptom-based approaches miss.
BEDTIME CHANNEL: NORMALIZED. 22:17 bedtime tonight is back into the pre-cluster-worst range. The 50-min-earlier bedtime + high compliance (7.80 h use, 7.65 h TIB) drove the exceptional sleep quality metrics. Bedtime discipline appears to be working; consider making 22:15–22:30 the default target.
What to watch on 08-22. (1) AHI/OA — if the 4-night escalation extends to 5 (AHI ≥ 2.82, or OA ≥ 15), the cluster is deeply entrenched and clinician contact should not wait for the next scheduled check-in; if AHI drops back below 2.0, the escalation may have hit its ceiling and the peak is 08-21. Either outcome is informative. (2) Leak avg — needs to stay ≤ 0.5 to close the 3-night mask-fix confirmation. (3) P_90 — should stay ≤ 10.5 if tonight's read is the new normal; if P_90 climbs while AHI stays elevated, that would flag a shift toward sustained obstruction. (4) TIB and bedtime — 22:15–22:30 window preferred; anything past 23:00 will cost total sleep. (5) H count — tonight's 6 was unusually high; a return to typical 2–3 would be reassuring, sustained ≥ 5 would indicate the event mix is shifting.
2026-08-21T22:17:53) marks the start calendar date, which lines up with the OSCAR "Period" for 2026-08-21. This convention continues to hold consistently.