Sleep & CPAP Report — 2026-10-08

🚩 Flags

• 🟢 CPAP metrics are the cleanest of the recent window — second-lowest AHI, zero hypopneas, and the lowest pressures in 26 nights. AHI 0.85 (rank 2/26, i.e. second-lowest in the recent 26-night window — only 10-01 (0.56) was lower; 30d z = -1.33, -58.5% vs 30d mean of 2.05). OA count 6 (rank 7/26, z = -0.83, -35.0% vs 30d); H count 0 (rank 2/26 tied — only 10-01 also had H=0 in the window; z = -1.22); CA 0 (unchanged, 7th straight night without a central); RERA 1 (vs 0 yesterday — negligible). Tonight is the third consecutive clean follow-up to the 10-05 breakthrough (AHI 1.18 → 5.20 → 1.85 → 1.24 → 0.85) — any argument that 10-05 was the start of a trend is now firmly closed.

• 🟢 Auto-titrator ran at the low end of the user's operating range — lowest pressures in 26 nights. Pressure_Avg 6.89 cm H₂O (rank 2/26, only 10-01 (6.82) was lower; 30d z = -1.51, -8.7% vs 30d mean of 7.55); Pressure_90 9.20 (rank 2/26, z = -1.40, -9.2% vs 30d mean of 10.13); Pressure_Max 11.50 (vs yesterday's 12.62, -8.9%). The device needed almost no pressure escalation tonight — consistent with the near-zero event count.

• 🔴 Deep sleep dropped sharply despite the pristine CPAP night — 9.40% is the second-lowest in the recent 7-night window and 5th-lowest in 26 nights. Deep 9.40% / 38.6 min (rank 5/26; 30d z = -1.15, -30.7% vs 30d mean of 13.56%). This is a decoupled signal: CPAP delivered its best night in weeks, but sleep architecture did not reward it with deep sleep gains. Partial mean-reversion is expected after yesterday's 18.21% peak, but the drop is larger than a simple regression — 9.40% is below both the 7d (13.14) and 30d (13.56) means by substantial margins. REM 24.30% / 99.9 min (z = -0.01, in-line with 30d); awake 9.03 min (in-line, z = -0.08).

• 🔴 Leak_Max streak extends to 7 straight nights above 22.8 L/min — the Leak_Max pattern is no longer an occasional outlier. Leak_Max 30.00 L/min (rank 22/26, z = +0.98, +39.4% vs 30d mean of 21.52). Every night since 2026-10-02 has had Leak_Max ≥ 22.8 L/min (25.2, 22.8, 28.8, 22.8, 30.0, 34.8, 30.0 → mean of 27.77 for the trailing 7d, +23.7% vs the prior 7d's 22.46). Leak_Avg 1.58 L/min is also elevated (z = +0.69, +46.6% vs 30d mean of 1.08). Leak_90 3.6 (z = +0.06 — modest, consistent with brief transients rather than sustained drift). What was flagged as a 4-night pattern yesterday is now a 7-consecutive-night pattern — mask fit / cushion wear / sleep position are overdue for a check. Important nuance: tonight's leak transients did not translate into event-count elevation (AHI 0.85) or pressure escalation, so the auto-titrator is still handling the seal problem gracefully — but the trend is sustained.

• 🟢 Flow-limitation metrics reverted to baseline after yesterday's elevation — the FL spike was not a trend. FL_P95 0.09 (z = -0.23, in-line with 30d mean of 0.094 — down from yesterday's 0.12); Min_Above_010 17.4 min (z = +0.17, essentially on the 30d mean of 16.64; down from yesterday's 23.3); Min_Above_020 3.9 min (z = -0.02, on the 30d mean). Frac_NonZero 22.13% (down from yesterday's 34.94%). Everything FL-related reset to normal in a single night — reinforces that yesterday's FL elevation was a one-off.

• 🟢 Longest sleep night in the recent window — CPAP usage 7.08 hours, TIB 420 min, TSD 411 min. CPAP Hours 7.08 (rank 24/26, z = +1.65, +10.5% vs 30d mean of 6.41 — only 09-16 (7.72) and 09-26 (6.98) came close); TIB 420.0 min (z = +1.58, +10.7% vs 30d mean of 379.4); TSD 411.0 min (z = +1.61, +11.1% vs 30d mean of 369.9). Sleep efficiency 97.85%. Despite the lower deep%, total sleep time was outstanding — the user stayed in bed longer than any night in the window.

• 🟡 7d AHI avg climbed to 2.10 (+27.2% WoW), still 10-05-driven. The 10-05 outlier (AHI 5.20) is still inside the recent 7d window. Drop it and the 7d avg is ~1.57 — well below the prior 7d's 1.65 and in line with the user's recovery trajectory. The headline WoW figure is almost entirely a single-night artifact.

• 🟡 Deep% is trending down WoW — 13.14 vs 15.19 prior 7d (-13.5%). This is driven by the four poor-deep nights in the recent 7d window (10-03 9.63%, 10-04 7.88%, 10-06 12.12%, 10-08 9.40%) partially offset by 10-07's 18.21% and 10-02's 18.86%. REM% is up WoW (+20.8%) and awake min is down (-48.6%), so overall architecture is still improving — but deep specifically is in a soft patch worth watching for another night or two.

Summary Table — Night of 2026-10-08

Metric	Value
Sleep start	2026-10-08T22:14:34-05:00
Time in bed	420.00 min (7h 0.0m)
Total sleep	410.97 min (6h 51.0m)
Sleep efficiency	97.85%
Awake time	9.03 min
Core (N1+N2) sleep	272.5 min / 66.30%
Deep (N3) sleep	38.64 min / 9.40%
REM	99.9 min / 24.30%
CPAP usage hours	7.08
Session count	1
AHI	0.85
RDI	0.99
Obstructive apneas (OA)	6
Hypopneas (H)	0
Central apneas (CA)	0
Unclassified apneas (UA)	0
RERA	1
Pressure_Avg	6.89
Pressure_Min	4.00
Pressure_Max	11.50
Pressure_90th	9.20
Leak_Avg	1.58
Leak_Max	30.00
Leak_90th	3.60
FL_Avg	0.0141
FL_P95	0.09
FL_Max	0.54
Min_Above_005 (min)	37.2
Min_Above_010 (min)	17.4
Min_Above_020 (min)	3.9
Frac_NonZero (FL samples)	22.13%

Comparison to Trailing Averages

Metric	Tonight	7d avg	30d avg	Δ vs 7d	Δ vs 30d
AHI	0.85	2.10	2.05	-59.5%	-58.5%
OA count	6.00	9.14	9.23	-34.4%	-35.0%
H count	0.00	3.86	3.13	-100.0%	-100.0%
CA count	0.00	0.29	0.63	-100.0%	-100.0%
RERA count	1.00	2.00	0.90	-50.0%	+11.1%
Pressure_Avg	6.89	7.62	7.55	-9.5%	-8.7%
Pressure_90	9.20	10.33	10.13	-11.0%	-9.2%
Leak_Avg	1.58	1.75	1.08	-9.6%	+46.6%
Leak_Max	30.00	27.77	21.52	+8.0%	+39.4%
Leak_90th	3.60	5.66	3.40	-36.4%	+5.9%
Usage hours	7.08	6.48	6.41	+9.3%	+10.5%
FL_P95	0.090	0.097	0.094	-7.4%	-4.6%
Min_Above_010	17.40	17.77	16.64	-2.1%	+4.6%
Min_Above_020	3.90	4.81	3.93	-19.0%	-0.8%
FL_Avg	0.0141	0.0168	0.0156	-15.9%	-9.8%
Frac_NonZero	0.221	0.261	0.252	-15.2%	-12.2%
Deep %	9.40	13.14	13.56	-28.5%	-30.7%
REM %	24.30	26.82	24.32	-9.4%	-0.1%
Awake min	9.03	6.81	9.52	+32.6%	-5.1%
TIB min	420.00	387.37	379.42	+8.4%	+10.7%
Total sleep min	410.97	380.56	369.90	+8.0%	+11.1%

Week-over-Week Deltas (recent 7d 2026-10-02–2026-10-08 vs prior 7d 2026-09-25–2026-10-01)

Metric	Recent 7d	Prior 7d	Δ	Δ %
AHI	2.099	1.650	+0.449	+27.2%
OA	9.143	8.000	+1.143	+14.3%
H	3.857	2.429	+1.429	+58.8%
CA	0.286	0.143	+0.143	+100.0%
Pressure_Avg	7.616	7.701	-0.086	-1.1%
Pressure_90	10.331	10.326	+0.006	+0.1%
Leak_Avg	1.749	1.106	+0.643	+58.1%
Leak_90	5.657	3.943	+1.714	+43.5%
Leak_Max	27.771	22.457	+5.314	+23.7%
Usage hours	6.477	6.466	+0.011	+0.2%
Deep % (Apple)	13.144	15.191	-2.046	-13.5%
REM % (Apple)	26.824	22.210	+4.614	+20.8%
Awake min (Apple)	6.810	13.261	-6.452	-48.6%
Sleep min (Apple)	380.564	369.244	+11.320	+3.1%
TIB min (Apple)	387.374	382.506	+4.868	+1.3%
FL_P95	0.097	0.094	+0.003	+3.0%
Min_Above_010	17.771	16.571	+1.200	+7.2%
Min_Above_020	4.814	3.186	+1.629	+51.1%

Recent 10-Night Context

Date	AHI	OA	CA	PAvg	P90	LAvg	LMax	L90	FL_P95	MinAbv010	MinAbv020	Deep%	REM%	Awake
2026-09-29	1.45	8	0	8.10	11.34	0.44	33.6	0.0	0.07	10.6	2.5	13.30	15.99	21.08
2026-09-30	1.44	7	0	8.01	10.40	1.18	16.8	3.6	0.10	16.9	2.0	16.10	18.32	24.59
2026-10-01	0.56	4	0	6.82	9.00	0.53	8.4	2.4	0.07	10.9	2.2	16.96	22.96	16.56
2026-10-02	2.04	9	1	8.50	10.32	1.68	25.2	8.4	0.09	15.9	2.8	18.86	34.36	8.53
2026-10-03	2.33	7	1	7.21	9.30	2.77	22.8	10.8	0.08	11.3	2.5	9.63	25.46	10.54
2026-10-04	1.18	6	0	7.90	10.18	1.91	28.8	4.8	0.13	25.7	7.1	7.88	27.87	1.51
2026-10-05	5.20	21	0	7.93	12.36	1.76	22.8	7.2	0.09	15.7	8.2	15.92	25.21	8.03
2026-10-06	1.85	8	0	7.19	10.24	1.04	30.0	1.2	0.08	15.1	4.6	12.12	27.14	4.01
2026-10-07	1.24	7	0	7.69	10.72	1.50	34.8	3.6	0.12	23.3	4.6	18.21	23.43	6.02
2026-10-08	0.85	6	0	6.89	9.20	1.58	30.0	3.6	0.09	17.4	3.9	9.40	24.30	9.03

CPAP → Sleep-Stage Correlations (all 100 matched nights)

Correlations computed across full available history (2026-06-23 → 2026-10-08).

Pair	the user r	Interpretation
AHI vs deep_pct	+0.069	negligible
AHI vs awake_min	+0.039	negligible
OA vs awake_min	+0.093	negligible
CA vs deep_pct	+0.054	negligible
Leak_Avg vs deep_pct	-0.142	weak negative
Leak_90 vs deep_pct	-0.111	weak negative
Leak_Max vs awake_min	+0.026	negligible
FL_P95 vs awake_min	-0.080	negligible
Min_Above_010 vs deep_pct	+0.019	negligible

Bucket comparison (Leak_90 median split, median = 2.4): - Low-leak nights (n=61): mean deep% 15.20, awake 14.34 min - High-leak nights (n=39): mean deep% 13.59, awake 11.53 min - Δ (high – low): deep% -1.61 pp, awake -2.81 min

The direction remains consistent with earlier reports: higher 90th-percentile leak is weakly associated with less deep sleep. Effect size is small (~1.6 percentage points of deep sleep on average on high-leak nights) and has been stable across weeks of data. Tonight is a case where the bucket signal does align weakly with the night's data (Leak_90 3.6 is just above the median → mild "high-leak" bucket, deep% 9.40 is low) — but the effect size seen tonight (~4 pp below the "high-leak" bucket mean) is much larger than the correlation would predict, meaning other factors are likely in play beyond leak alone.

Caveats

• Tonight (10-08) is the third consecutive clean follow-up after the 10-05 breakthrough. The AHI 5.20 spike on 10-05 is now firmly surrounded by five consecutive clean event-count nights (10-04 1.18, 10-06 1.85, 10-07 1.24, 10-08 0.85) and is confidently a one-off.
• Leak_Max has been ≥ 22.8 L/min for 7 consecutive nights now (10-02 through 10-08). This is a sustained pattern, not an occasional outlier. The 7d-vs-prior-7d Leak_Max delta is +23.7%. Mask fit, cushion wear, headgear tension, or sleep position are overdue for inspection. The auto-titrator is still handling the seal problem gracefully (AHI stayed near zero tonight despite Leak_Max 30), so this is a maintenance signal, not a clinical one — but it is now trend data, not noise.
• Deep sleep is in a soft patch — 4 of the 7 recent-week nights were below 13% (10-03, 10-04, 10-06, 10-08). WoW deep% is -13.5%. This is worth watching across the next 2–3 nights to see whether it reverts to the ~15% baseline or stays depressed.
• Tonight's CPAP-vs-architecture decoupling is noteworthy: AHI 0.85 and pressures at the window's low end but deep% at 9.40%. Possible mean-reversion from yesterday's 18.21% peak; possible positional / environmental cause. Correlations are too weak to attribute to a single CPAP channel. Note that tonight also had the longest TIB (420 min) and longest TSD (411 min) in the window — the body stayed in bed longer, but the extra time was core, not deep.
• Flow-limitation elevation flagged yesterday did not persist — FL_P95, Min_Above_010, Min_Above_020, and Frac_NonZero all returned to the 30d baseline tonight. Yesterday's FL elevation can be treated as a single-night deviation.
• Matched-night count is 100 (one more than yesterday — 10-08 joined the matched set after both exports became available this morning). This crossed the 100-matched-night milestone.
• Early 2026-06 nights (via multi-day roll-up) retain zero-valued pressure/leak channels and remain low-confidence for CPAP metrics prior to ~2026-06-22.
• Correlations are directional/small-effect only — single-night deviations often exceed what correlations alone would predict (tonight is a clear example — CPAP metrics the best in weeks but deep% near the bottom of the window).
• All flagged metrics for tonight remain within clinically safe ranges (AHI < 5, leak within mask spec). Flags mark deviations from the user's own recent baseline and connections to recent trends, not clinical thresholds.