Daily Report — 2026-08-19

Sleep & CPAP Report — Night of 2026-08-19

Sources: health-report-2026-08-19-to-2026-08-20.json (Apple Watch sleep stages, exported 05:26 on 08-20) + OSCAR_Greg_by_Day_2026-08-19.csv (ResMed AirSense/AirFlow 11 via OSCAR). Session start was 22:28 on 08-19 — 41 minutes earlier than 08-18's 23:09, breaking the 3-night 23:02–23:09 cluster with the earliest bedtime in the trailing 10-night window. CPAP hours 6.83 vs. Watch TIB 6.71 (Δ ≈ 7.2 min, CPAP slightly longer — the machine was on for a brief awake tail after Watch closed the session; tight overall alignment).

⚠️ Flags

MASK-FIT ACTION TRIGGERED — 3rd consecutive night with leak avg ≥ 1.0, and tonight's reading jumped +61% to 1.92 L/min. The 5-night leak avg trajectory is now 0.31 → 0.91 → 1.14 → 1.19 → 1.92 L/min (08-15 → 08-19). Tonight is the 3rd consecutive ≥1.0 night, crossing the threshold set in the 08-17 report ("3 consecutive ≥1.0 warrants a physical mask/strap/cushion check"). This is no longer plausibly a "self-correcting fit adjustment" — the trend is monotonically accelerating, not stabilizing. Leak_90 climbed 3.6 → 4.8 L/min (+33%), leak max 22.8 → 26.4 L/min. Recommend: strap tension, cushion seat, and pillow/nasal seal check before tonight (08-20). Also worth inspecting the cushion silicone for a tear or deformation given 5 nights of monotonic worsening — a mechanical failure fits this signature better than positional drift.
AHI CLUSTER REOPENED — OA count = 10 hit the reopen threshold exactly. The 08-17 report set the disambiguation: "AHI ≥ 2.5 or OA ≥ 10 = cluster reopens." Tonight: AHI 2.20 (below the 2.5 line but the highest since 08-12's 2.68) and OA 10 (hits threshold). 3-night OA trajectory: 5 → 7 → 10 — monotonically rising, +100% over three nights. 3-night OA/h rate for 08-17→08-19: 22 OA / 19.23 h use = 1.14 /h, matching the 30-day baseline exactly, but the vector is clearly wrong. The 08-11 → 08-15 elevated period is now extended, not resolved. Note: the leak-worsening and AHI-rebounding are running in parallel — a compromised mask seal typically raises event counts as the machine's ability to deliver stable pressure degrades. These two flags are likely linked, not independent.
Auto-titrator raised its baseline — P_avg jumped +0.63 cmH₂O in one night to 8.18, the highest P_avg of matched history (n=32). Prior 08-18 P_avg was 7.55; the 30-day baseline is 7.44. Tonight's 8.18 exceeds every night in the matched window. Meanwhile P_90 pulled back slightly (11.11 → 10.34) but stayed above 7-day (10.34 vs. 10.18) and 30-day (10.00). Interpretation: the machine is shifting from event-driven pressure spikes toward a persistently higher baseline — this is the algorithm treating the load as ongoing rather than episodic. Combined with the leak trend, the parsimonious reading is: degraded seal → higher effective demand → sustained higher baseline pressure.
Sleep architecture RECOVERED — closes the 08-18 concern. Deep 16.50% (above 30-day 14.56%, +1.94 pts), REM 24.11% (essentially at 30-day 24.09%). This resolves the 08-18 report's contingent question — "if 08-19 also comes in with Deep < 12%, treat as a 3-night architecture regime." Tonight's 16.50% deep clearly rejects that; 08-18 was a one-night noise event on top of the 08-13/14 dip, not a new regime. Combined with tonight's very low awake time (7.0 min, rank 5/32), this is the best-consolidated sleep of the last four nights. Notably, this recovery happened despite the worsening CPAP-side signals (higher leak, higher AHI, higher pressure), reinforcing the persistent finding that architecture and CPAP metrics are decoupled in this dataset.
Bedtime shift — 22:28 is the earliest bedtime in the last 10 nights. Whether the earlier bedtime contributed to the strong architecture recovery is not identifiable from a single night, but note the association: earliest bedtime + longest TIB (6.71 h) + top-5 lowest awake time + top-2 highest efficiency (98.25%) + normal deep/REM. If tonight's earlier bedtime was intentional, worth continuing to see if the pattern repeats.

1. Summary Table — Night of 2026-08-19

Metric	Value
Bedtime (session start)	22:28 on 08-19 (41 min earlier than 08-18; earliest in trailing 10 nights)
Time in bed	6.71 h (402.4 min)
Total sleep duration	6.59 h (395.4 min)
Sleep efficiency (TST/TIB)	98.25% — rank 2/32 (highest matched); +3.07 pts vs. 7-day (95.18%), +2.83 pts vs. 30-day (95.42%)
Awake time	7.02 min — rank 5/32 lowest; −64% vs. 7-day (19.5), −63% vs. 30-day (18.8)
Core sleep	3.91 h (234.8 min) — 59.39% (back into typical 55–65% band)
Deep sleep	1.09 h (65.2 min) — 16.50% — +7.64 pts vs. 08-18 (8.86%); +1.94 pts vs. 30-day (14.56%)
REM sleep	1.59 h (95.3 min) — 24.11% — +5.40 pts vs. 08-18 (18.71%); essentially at 30-day (24.09%)
CPAP usage	6.83 h — +11.6% vs. 08-18 (6.12); +2.1% vs. 7-day (6.69), +1.8% vs. 30-day (6.71)
AHI	2.20 — rank 27/32 (near-top); +35% vs. 7-day (1.63), +34% vs. 30-day (1.64); +35% vs. 08-18 (1.63)
RDI	2.20 (RERA = 0)
OA / H / CA / UA / RERA	10 / 5 / 0 / 0 / 0 — OA hits cluster-reopen threshold; CA/RERA both clean
OA per hour	1.46 /h — +28% vs. 30-day (1.14/h); +28% vs. 08-18 (1.14/h)
Pressure avg	8.18 cmH₂O — top of matched history (rank 32/32); +0.63 vs. 08-18 (7.55); +0.46 vs. 7-day (7.72); +0.74 vs. 30-day (7.44)
Pressure 90th %ile	10.34 cmH₂O — pulled back from 08-18's 11.11 but still +0.16 vs. 7-day (10.18), +0.34 vs. 30-day (10.00)
Leak avg	1.92 L/min — 5th consecutive rising night, 3rd consecutive ≥1.0; +61% vs. 08-18 (1.19); +217% vs. 7-day (0.61); +140% vs. 30-day (0.80)
Leak 90th %ile	4.8 L/min — +33% vs. 08-18 (3.6); rank 28/32 (near-top of matched history)
Leak max	26.4 L/min
Compliance flag	1 (compliant)

2. 10-Night Trend & Context

Date	AHI	OA	H	CA	RERA	Leak avg	Leak 90	P_avg	P_90	Hours	Deep %	REM %	Awake (min)	Eff
08-09	1.34	6	3	0	2	3.00	10.8	7.79	10.94	6.72	10.97	23.65	15.1	96.19%
08-10	0.58	3	0	0	0	0.17	0.0	6.89	9.46	5.20	15.30	20.31	20.1	94.73%
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	13.6	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%

(08-15 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.)

Four-night composite (08-16 → 08-19): AHI mean 1.52 • OA total 25 across 25.45 h use = 0.98 OA/h • Awake total 34.1 min across 25.42 h TIB = 2.24% awake fraction • Deep mean 13.58% • REM mean 21.99% • Efficiency mean 97.72%. Compared to the 3-night composite through 08-18 (Deep 12.60%, REM 21.28%, AHI 1.29), tonight's addition pulls AHI up +0.23 but brings Deep and REM both back up. The sleep-quality picture is improving; the CPAP picture is getting worse.

Seven-night 08-13 → 08-19 obstructive composite: OA 52, H 12, CA 4, RERA 4 across 46.46 h = 1.12 OA/h. Compare to the 08-12 → 08-18 composite (59 OA in 46.35 h = 1.27 OA/h). Weekly OA/h rate ticked down slightly (only because 08-12's 13-OA night dropped out of the window and 08-19's 10 is smaller by comparison), but the within-week trajectory is monotonically worsening — the last three nights are 5 → 7 → 10.

Leak progression (last 6 CPAP nights, including 08-15): 0.31 → 0.91 → 1.14 → 1.19 → 1.92 L/min for the matched 5-night run 08-15 → 08-19. Monotonic. Rate of increase itself is accelerating: nightly deltas are +0.60, +0.23, +0.05, +0.73. Tonight's +0.73 is the largest single-night jump of the series and reverses the deceleration seen 08-16 → 08-18. This kills the "converging drift" reading.

Leak vs. sleep architecture — 32 matched nights (through 08-19)

Buckets on nightly leak average:

Low-leak (≤ 0.30 L/min): n = 7 — Mean deep: 13.27% | Mean awake: 23.3 min
Higher-leak (> 0.30 L/min): n = 25 — Mean deep: 14.76% | Mean awake: 17.3 min
Very-high-leak (≥ 1.0 L/min): n = 10 — Mean deep: 13.75% | Mean awake: 12.8 min

Tonight's data point (leak 1.92, deep 16.50%, awake 7.0 min) reinforces the persistent finding: high-leak nights do not predict poor sleep architecture in this dataset. The very-high-leak bucket now has 10 nights with mean deep 13.75% and mean awake time 12.8 min — actually the most consolidated sleep of the three buckets. the user correlations (n=32) essentially unchanged from prior report: leak avg vs. deep %: ~ −0.05; leak avg vs. awake min: ~ −0.30 (directional, weak); AHI vs. deep %: ~ +0.03; AHI vs. awake min: ~ −0.08; pressure 90 vs. deep %: ~ +0.05. The independence of the sleep-quality channel and the CPAP-mechanics channel is now a very robust finding in this dataset.

3. Comparison to Trailing Averages

Metric	Tonight (08-19)	7-day avg (through 08-18)	30-day avg (through 08-18)
AHI	2.20	1.63	1.64
OA count	10	7.86	7.47
OA per hour	1.46 /h	~1.19 /h	~1.14 /h
Leak avg (L/min)	1.92	0.61	0.80
Leak 90th (L/min)	4.8	~1.9	2.44
Pressure avg (cmH₂O)	8.18	7.76	7.44
Pressure 90th (cmH₂O)	10.34	10.34	10.00
Usage hours	6.83	6.62	6.71
Deep sleep %	16.50%	13.21%	14.56%
REM sleep %	24.11%	23.14%	24.09%
Awake time (min)	7.02	18.15	18.8
Efficiency	98.25%	95.54%	95.42%
TIB (h)	6.71	6.62	6.63

Verdict: the sleep-quality/CPAP-mechanics split is at its widest of the matched history. Sleep architecture, consolidation, and efficiency are all clearly better than baseline (deep +1.94 pts, awake −11.8 min, efficiency +2.8 pts). Every CPAP mechanics metric is clearly worse — AHI +34%, OA +34%, leak avg +140%, P_avg the highest of matched history. Interpreting either channel in isolation would give an opposite conclusion about the night. The reconcilable reading: the CPAP is working harder to compensate for a degraded seal, but the compensation is largely succeeding at maintaining sleep quality — for now. Continued leak degradation eventually erodes therapy effectiveness; tonight is not that point yet, but the trajectory says it could be within days.

Weekly AHI/Leak/Pressure/OA trend (CPAP-only, full week nights)

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 (partial, n=1)	2.20	1.92	8.18	10.34	10.0	6.83	1

The new week is opening with the highest single-night leak avg of the last 32 nights, the highest single-night P_avg of matched history, and OA count at the cluster-reopen threshold. This is the worst CPAP-mechanics night since at least 08-12 across leak, pressure, and event count simultaneously. Even 08-09 (leak avg 3.00) was worse only on the leak channel — AHI was 1.34 and P_avg 7.79, so tonight is more broadly elevated. The weekly rollup will need 6 more nights before a fair comparison, but the opening signal is negative.

4. Outlook

Priority action: mechanical mask check tonight. The 5-night monotonic leak trend, tonight's +61% single-night jump, and the parallel AHI/OA/pressure elevations all point to a seal integrity issue. The most useful diagnostic actions before 08-20 sleep: (1) inspect cushion silicone for tears, deformation, or discoloration — 5 nights of monotonically worsening leak fits a mechanical failure signature better than positional drift; (2) verify strap tension symmetry; (3) confirm the cushion is fully seated in the frame; (4) if using a pillows-style interface, check the pillow seals for splits. If tonight (08-20) shows leak avg back under 0.5 after the check, treat this as a diagnosed and fixed mechanical issue. If leak stays ≥1.0, the cushion probably needs replacement.

AHI cluster reopens. Per the 08-17 disambiguation thresholds, tonight's OA=10 hits the reopen line. The two-night improvement of 08-16/17 is now clearly the exception in a broader elevated window running from 08-11. Watch for two things on 08-20: (1) whether an actual mask fix drops leak → does AHI/OA fall proportionally, confirming the linkage; (2) whether AHI stays ≥2 even after a leak fix, which would suggest a real event-load elevation independent of therapy delivery. The natural experiment tomorrow: if fixing the leak also fixes the AHI, this whole cluster is a mechanical-therapy problem and not an underlying obstructive-load problem.

Sleep architecture concern from 08-18 is resolved. Deep 16.50% and REM 24.11% both back above 30-day baseline. Combined with excellent consolidation (98.25% efficiency, 7.0 min awake), 08-18 was single-night noise. No follow-through action needed on the architecture channel.

Pressure baseline shift is the second-most notable finding. P_avg 8.18 is the highest of matched history. The auto-titrator is clearly responding to sustained load, not one-off events. If a leak fix on 08-20 also drops P_avg back to ~7.5, that confirms the whole pressure elevation was a compensation for lost seal. If P_avg stays ≥8 even after a leak fix, the machine is seeing real sustained obstructive demand — worth mentioning to the sleep clinician on the next check-in.

What to watch on 08-20. (1) Leak avg — the single most important metric this cycle; ≤0.5 = fix confirmed; 0.5–1.0 = partial; ≥1.0 = mechanical issue persists; ≥1.9 (repeat of tonight) = cushion replacement likely needed. (2) AHI/OA — proportional recovery with leak = coupled system; independent AHI elevation = real event-load. (3) P_avg — return to ~7.5 = titrator-compensation confirmed; stay ≥8 = sustained real load. (4) Deep/REM % — expect continued baseline-normal readings; the architecture channel is not the concern this cycle.

5. Caveats

CPAP/Watch alignment tonight is very tight in the opposite direction from usual — CPAP hours 6.83 exceed Watch TIB 6.71 by 7.2 min, meaning the machine ran briefly past the end of the tracked sleep session (likely a short awake period at the tail). Per-hour rates and architecture comparisons remain directly interpretable.
The parallel movement of leak, AHI, OA, and P_avg is the single most information-dense signal tonight. Any of these individually could be noise; all four moving together in the same direction is signal — most parsimoniously a seal degradation causing therapy compensation. The interpretation could still be wrong if there's an independent respiratory-load driver (illness, positional, alcohol/sedatives, congestion) that coincidentally aligned with a fit issue, but the mechanical explanation is the leaner one.
Earlier bedtime (22:28) coinciding with excellent architecture recovery is suggestive but not proven. A single night cannot establish that earlier bedtime causes the improvement, since the 08-13/14/18 dips had different bedtimes and different other factors. Worth tracking if the earlier bedtime becomes a pattern.
P_avg 8.18 is the highest of matched history (n=32). Prior high was around 8.35 on the 08-15 CPAP-only night (which is not in the matched-history denominator). Within matched nights, tonight is the new high.
OA=10 exactly hits the reopen threshold. The 08-17 report set that as a reopen trigger (≥10), so tonight is inside the trigger zone by one event. A single event's worth of ambiguity.
Correlations still show no meaningful CPAP-vs-architecture relationships in n=32. Tonight's high-leak/high-deep combination continues the pattern of independence between these two channels. Only the leak-vs-awake correlation (~−0.30) has any directional signal, and it goes the "wrong" way (higher leak associates with lower awake time, not higher).
CPAP-nights CSV format. Newer files include 90th-percentile leak; older files use 95th-percentile. Both treated as the same "tail leak" metric for trend purposes.
No SpO2 data in tonight's CPAP CSV export (columns present but blank) — consistent with every export in the current format.
Matched-night count on this run: n=32. CPAP-only nights (no Watch match) in the last 6 weeks remain: 07-17, 08-07, 08-15. Weekly CPAP-only rollups use full n=6–7; matched-history stats use n=32.
Session-start convention. The date field of the sleep session (2026-08-19T22:28:06) marks the start calendar date, which lines up with the OSCAR "Period" for 2026-08-19. This convention has held consistently through the matched-history window.

See all past reports →