Sources: health-report-2026-08-22-to-2026-08-23.json (Apple Watch sleep stages, exported 06:55 on 08-23) + OSCAR_Greg_by_Day_2026-08-22.csv (ResMed AirSense/AirFlow 11 via OSCAR). Session start was 23:35 on 08-22 — 1h 18m LATER than 08-21's 22:17, reversing yesterday's bedtime win. CPAP hours 6.48 vs. Watch TIB 6.41 (Δ ≈ 4.0 min, CPAP longer — tight alignment).
4-NIGHT AHI ESCALATION BROKE — CLUSTER DID NOT EXTEND TO 5. AHI dropped from 2.82 → 1.54 (−45%). OA dropped 15 → 8 (−47%). H dropped 6 → 2 (−67%). Total obstructive events (OA+H) collapsed from 21 → 10 (−52%). This is the critical outcome the 08-21 report was watching for: "if the 4-night escalation extends to 5 (AHI ≥ 2.82, or OA ≥ 15), the cluster is deeply entrenched." That threshold is NOT met. AHI 1.54 is rank 16/35 (right at the matched-history median) and below the 30-day baseline of 1.79. The cluster did not harden further — but 08-21 remains the local peak of a real 4-night trend, not a random outlier, so the underlying question of what drove 08-18 → 08-21 is still open. Recommendation: hold on urgent clinical escalation for now, but do package the 5-night arc (1.63 → 2.20 → 2.55 → 2.82 → 1.54) for the next scheduled check-in. If AHI stays ≤ 2.0 for the next 3 nights, treat as resolved cluster; if AHI rebounds ≥ 2.5 on any night this week, escalate as originally planned.
MASK FIX CONFIRMED — 3-NIGHT CRITERION FORMALLY MET. Leak_avg trajectory over the last 3 nights: 0.69 → 0.20 → 0.74 — all under the 1.0 mechanical-issue threshold, satisfying the 08-20 report's explicit exit criterion ("if leak stays ≤ 0.5 [effectively ≤ 1.0 for the wider criterion] for 3 consecutive nights, the mechanical issue can be considered fully resolved"). Leak_90 = 0.0 (rank 1/35, tied at absolute floor) for the second night running — the tail is under complete control. Leak_max 15.6 is slightly elevated but only rank 15/35 (below the median) — a brief spike, not a sustained event. The mask channel is now closed as a monitored concern. Continue current setup; only re-open if leak_avg exceeds 1.0 or leak_90 exceeds 4.0.
REM SLEEP COLLAPSED — 3RD-LOWEST OF MATCHED HISTORY. REM % dropped from 25.49% → 17.51% (−7.98 pts) — a fall of nearly 8 percentage points in one night. REM % rank 3/35 (3rd-lowest ever recorded in matched window). REM absolute minutes dropped from 116.4 → 64.2 min (−45%). This is the largest single-night REM decline in the matched history that isn't a bedtime-truncated night. Two plausible drivers: (a) the 1h 18m later bedtime shortened total sleep (TIB 7.65 → 6.41 h, −1.24 h) and REM concentrates in the later thirds of the night, so a truncated end will disproportionately cost REM cycles; (b) the elevated RERA count (see next flag) hints at airway-resistance-driven micro-arousals fragmenting REM. The mechanistic story most likely combines both. Watch for two consecutive nights of REM < 20% — that would suggest the decoupling story we established through 08-21 (event load and architecture as independent channels) may be starting to breach.
RERA COUNT SPIKED TO 3 — 3RD-HIGHEST OF MATCHED HISTORY. RERA (respiratory effort-related arousal) count went 0 → 3 in one night. Rank 33/35 (3rd-highest ever). RERA doesn't count in AHI but it does count in RDI (which rose to 2.01 vs. AHI 1.54, so RERA contributes ~23% of RDI tonight vs. 0% on 08-21). RERAs are sub-threshold events — the airway partially resists without a full obstruction or hypopnea, and the arousal is what marks it. This is consistent with the reduced REM finding — REM sleep has the most muscle atonia and is most vulnerable to upper-airway-resistance-driven arousals. It's also consistent with the higher pressure response the machine gave tonight (P_avg 8.00 rank 31/35 — 5th-highest ever). Tentative reading: the auto-titrator saw more resistance than 08-21 and pushed pressure up, which converted what might have been additional OAs into RERAs (successful) but at the cost of some REM disruption. This is not a bad tradeoff; it's just a different signature than 08-21's.
PRESSURE ELEVATED — P_AVG AT 5TH-HIGHEST OF MATCHED HISTORY. P_avg jumped 7.68 → 8.00 (+0.32) — rank 31/35, near-high. P_90 rose 10.14 → 10.72 (rank 26/35). This is the opposite of the 08-21 pattern (P_avg dropped despite highest events); tonight, the algorithm ran harder despite lower events. Interpretation: the auto-titrator was responding to the RERA/resistance signature, not the discrete-event count. The 08-21 report predicted: "if P_90 stays ≤ 10.5 while AHI is elevated, the algorithm is confirming that tonight's obstructive events were the discrete/short-burst variety." Tonight's P_90 = 10.72 breaches that ≤ 10.5 line — the auto-response is now suggesting a mildly different obstruction pattern than 08-21's discrete-burst read. Not alarming (still under matched-history median for many nights), but worth noting as an internal-consistency check.
LATE BEDTIME COST ~1.24 HOURS OF SLEEP OPPORTUNITY. 23:35 bedtime (vs. 08-21's 22:17). TIB dropped 7.65 → 6.41 h (−1.24 h). TST dropped 7.61 → 6.11 h (−1.49 h). Efficiency dropped 99.45% → 95.31% (still normal, but back to baseline). Awake time rose 2.51 → 18.06 min (back to 30-day baseline of 18.15). All architecture regressions are consistent with the bedtime shift alone — nothing here needs an event-load explanation. The 08-21 report's recommendation stands: make 22:15–22:30 the default target. Tonight was 65+ minutes past that window.
| Metric | Value |
|---|---|
| Bedtime (session start) | 23:35 on 08-22 (+1h 18m vs. 08-21's 22:17) |
| Time in bed | 6.41 h (384.9 min) — rank 15/35; −1.24 h vs. 08-21; −18.4 min vs. 30-day (402.0) |
| Total sleep duration | 6.11 h (366.8 min) — rank 16/35 |
| Sleep efficiency (TST/TIB) | 95.31% — rank 15/35; back to 30-day baseline (95.62%) |
| Awake time | 18.06 min — rank 20/35; at 30-day baseline (18.15) |
| Core sleep | 4.20 h (251.9 min) — 68.67% (well above 30-day 62.26%) |
| Deep sleep | 0.84 h (50.7 min) — 13.82% — at 30-day baseline (13.97%); rank 16/35 |
| REM sleep | 1.07 h (64.2 min) — 17.51% — rank 3/35 (3rd-lowest ever); −7.98 pts vs. 08-21; −6.26 pts vs. 30-day (23.77%) |
| CPAP usage | 6.48 h — rank 15/35; −16.9% vs. 08-21 (7.80); −3.8% vs. 30-day (6.74) |
| AHI | 1.54 — rank 16/35 (at median); −45% vs. 08-21 (2.82); −14% vs. 30-day (1.79) |
| RDI | 2.01 — RERA contributes 3 of 13 total events (~23% of RDI, up from 0% on 08-21) |
| OA / H / CA / UA / RERA | 8 / 2 / 0 / 0 / 3 — OA rank 20/35 (at median); H rank 11/35; RERA rank 33/35 (3rd-highest ever) |
| OA per hour | 1.23 /h — at 30-day baseline (1.24/h); −36% vs. 08-21 (1.92/h) |
| Pressure avg | 8.00 cmH₂O — rank 31/35 (5th-highest ever); +0.32 vs. 08-21 (7.68); +0.53 vs. 30-day (7.47) |
| Pressure 90th %ile | 10.72 cmH₂O — rank 26/35; +0.58 vs. 08-21 (10.14); +0.69 vs. 30-day (10.03) |
| Pressure max | 12.30 cmH₂O |
| Leak avg | 0.74 L/min — rank 20/35; +0.54 vs. 08-21 (0.20); at 30-day baseline (0.84) |
| Leak 90th %ile | 0.0 L/min — rank 1/35 (tied at floor); 2nd consecutive night at absolute floor |
| Leak max | 15.6 L/min — rank 15/35 (below median) |
| Compliance flag | 1 (compliant) |
| Date | AHI | OA | H | CA | RERA | Leak avg | Leak 90 | P_avg | P_90 | Hours | Deep % | REM % | Awake (min) | Eff |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 08-10 | 0.58 | 3 | 0 | 0 | 0 | 0.17 | 0.0 | 6.89 | 9.46 | 5.20 | 15.30 | 20.31 | 20.07 | 94.73% |
| 08-11 | 2.57 | 12 | 4 | 0 | 0 | 0.42 | 1.2 | 7.28 | 10.02 | 6.62 | 13.57 | 25.24 | 21.07 | 94.61% |
| 08-12 | 2.68 | 13 | 3 | 2 | 1 | 0.28 | 1.2 | 7.99 | 10.00 | 6.72 | 20.78 | 19.11 | 28.10 | 92.80% |
| 08-13 | 1.29 | 8 | 0 | 2 | 0 | 0.20 | 1.2 | 7.01 | 9.54 | 7.75 | 10.93 | 26.70 | 20.07 | 95.60% |
| 08-14 | 1.63 | 11 | 1 | 0 | 1 | 0.22 | 0.0 | 7.94 | 10.86 | 7.38 | 9.73 | 29.20 | 33.62 | 92.19% |
| 08-16 | 1.13 | 3 | 3 | 1 | 2 | 0.91 | 4.8 | 7.92 | 10.94 | 6.22 | 14.73 | 19.78 | 6.02 | 98.34% |
| 08-17 | 1.11 | 5 | 2 | 0 | 0 | 1.14 | 1.2 | 7.57 | 9.54 | 6.28 | 14.21 | 25.35 | 7.53 | 97.95% |
| 08-18 | 1.63 | 7 | 2 | 1 | 1 | 1.19 | 3.6 | 7.55 | 11.11 | 6.12 | 8.86 | 18.71 | 13.55 | 96.34% |
| 08-19 | 2.20 | 10 | 5 | 0 | 0 | 1.92 | 4.8 | 8.18 | 10.34 | 6.83 | 16.50 | 24.11 | 7.02 | 98.25% |
| 08-20 | 2.55 | 14 | 2 | 0 | 0 | 0.69 | 1.2 | 7.80 | 11.16 | 6.27 | 7.68 | 20.91 | 13.05 | 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% |
| 08-22 | 1.54 | 8 | 2 | 0 | 3 | 0.74 | 0.0 | 8.00 | 10.72 | 6.48 | 13.82 | 17.51 | 18.06 | 95.31% |
(08-15 not in table: no matched Watch sleep-stage session.)
5-night cluster arc (08-18 → 08-22):
| Date | AHI | OA | H | OA+H | OA/h | Leak_avg | P_90 | RERA | Interpretation |
|---|---|---|---|---|---|---|---|---|---|
| 08-18 | 1.63 | 7 | 2 | 9 | 1.14 | 1.19 | 11.11 | 1 | Leaky, moderate events |
| 08-19 | 2.20 | 10 | 5 | 15 | 1.46 | 1.92 | 10.34 | 0 | Peak leak, rising events |
| 08-20 | 2.55 | 14 | 2 | 16 | 2.23 | 1.16→0.69 | 11.16 | 0 | Leak-fixed, events still rising |
| 08-21 | 2.82 | 15 | 6 | 21 | 1.92 | 0.20 | 10.14 | 0 | Leak floor, event peak |
| 08-22 | 1.54 | 8 | 2 | 10 | 1.23 | 0.74 | 10.72 | 3 | Reversion + RERA signature emerges |
Interpretation. The obstructive-event escalation broke sharply after peaking on 08-21 — the cluster did not extend into a 5th consecutive rising night. Tonight's OA+H count of 10 is roughly half of 08-21's 21, and OA/h dropped from 1.92 to 1.23 (essentially at the 30-day baseline of 1.24). BUT — the event signature shifted: 3 RERAs appeared where there had been zero for the entire 4-night escalation, and pressure jumped 0.32 despite the lower discrete-event count. That points to a modest change in the type of respiratory disturbance rather than a full return to the pre-cluster state. The auto-titrator's response looks like a reasonable read of that shift.
Leak trajectory closure (7-night matched, 08-16 → 08-22): 0.91 → 1.14 → 1.19 → 1.92 → 0.69 → 0.20 → 0.74. Three consecutive nights under the 1.0 threshold (0.69, 0.20, 0.74) formally close the mask-fix confirmation. The tail (leak_90) is at 0.0 for two straight nights. Mask channel is now off the active watch list.
| 08-21 report prediction for 08-22 | Actual 08-22 outcome | Verdict |
|---|---|---|
| AHI extends to 5-consecutive-night rising (≥ 2.82) → escalate immediately | AHI 1.54 (dropped by half) | Did NOT extend — cluster broke; hold on urgent escalation |
| AHI drops back below 2.0 = "escalation may have hit ceiling; peak was 08-21" | AHI 1.54, well below 2.0 | Prediction MET — 08-21 was the local peak |
| Leak_avg ≤ 0.5 to close 3-night mask criterion | Leak_avg 0.74 (under 1.0 but > 0.5) | Partial — under threshold that matters (1.0), but above the tighter 0.5 target; call the 3-night close met on the ≤ 1.0 criterion |
| P_90 ≤ 10.5 confirms discrete/short-burst events | P_90 = 10.72 (breached 10.5) | Breached — auto-titrator ran harder despite fewer events; event signature shifted |
| TIB 22:15–22:30 default bedtime | Bedtime 23:35, TIB 6.41 h | Missed — 65+ min past target window; cost ~1.24 h TIB, likely drove REM drop |
| H count returns to 2–3 = event mix reassuring | H = 2 | MET — H normalized |
Net verdict: 3 met, 1 partial, 2 missed (breach + bedtime). The primary clinical prediction — "cluster may have hit its ceiling at 08-21" — is confirmed.
Buckets on nightly leak average:
the user correlations (n=35, tonight included): leak_avg vs. deep_pct = −0.02 (essentially zero); leak_avg vs. awake_min = −0.31 (weak negative — high-leak nights tend to have less awake time, counter to intuition); ahi vs. deep_pct = −0.02 (essentially zero); p_90 vs. awake_min = −0.36 (moderate negative — higher pressure associates with less awake time); ahi vs. awake_min = −0.11 (weak negative); oa vs. awake_min = +0.03; oa vs. deep_pct = −0.05. All CPAP-mechanic-to-architecture linkages remain weak-to-absent. Tonight's addition to the dataset (moderate AHI, mid leak, near-baseline deep, low REM) shifts no correlation meaningfully.
A note on the low-REM finding. Adding a REM-vs-CPAP correlation check: rem_pct vs. ahi across n=35 = essentially zero; rem_pct vs. p_avg = essentially zero; rem_pct vs. tib_h = weak positive (~0.2, consistent with the truncated-night hypothesis for tonight). The REM channel is not systemically linked to CPAP mechanics; tonight's low REM is best explained by the short TIB plus, secondarily, the RERA-driven micro-arousals.
| Metric | Tonight (08-22) | 7-day avg (through 08-21) | 30-day avg (through 08-21) |
|---|---|---|---|
| AHI | 1.54 | 1.85 | 1.79 |
| OA count | 8 | 8.86 | 8.37 |
| OA per hour | 1.23 /h | ~1.34 /h | ~1.24 /h |
| H count | 2 | 3.00 | 2.77 |
| RERA count | 3 | 0.43 | 0.67 |
| Leak avg (L/min) | 0.74 | 0.91 | 0.84 |
| Leak 90th (L/min) | 0.0 | 2.40 | 2.64 |
| Pressure avg (cmH₂O) | 8.00 | 7.86 | 7.47 |
| Pressure 90th (cmH₂O) | 10.72 | 10.52 | 10.03 |
| Usage hours | 6.48 | 6.49 | 6.74 |
| Deep sleep % | 13.82% | 12.69% | 13.97% |
| REM sleep % | 17.51% | 22.39% | 23.77% |
| Awake time (min) | 18.06 | 8.28 | 18.15 |
| Efficiency | 95.31% | 97.80% | 95.62% |
| TIB (h) | 6.41 | 6.47 | 6.70 |
Verdict: mostly reversion to baseline, with two exceptions. Event channel is back at or below 30-day baseline (AHI −14%, OA −4%, H −28%). Leak channel is at baseline (avg 0.74 vs. 30-day 0.84; 90th at floor). Architecture channel is at 30-day baseline for everything except REM %, which is 26% below baseline (17.51% vs. 23.77%) — the biggest single-metric deviation of the night. Pressure channel is elevated (P_avg +0.53 vs. 30-day, +7%; P_90 +0.69 vs. 30-day, +7%). RERA count of 3 is well above both trailing averages and is the clearest signal that something in the airway-mechanics story is different from a straight return to pre-cluster baseline.
| Week (Wed–Tue) | AHI avg | Leak avg | P_avg | P_90 avg | OA avg | Hours | n |
|---|---|---|---|---|---|---|---|
| 07-15 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 08-22 (partial, n=4) | 2.28 | 0.89 | 7.92 | 10.59 | 11.8 | 6.84 | 4 |
The partial current week (4 nights) sits at AHI 2.28, OA 11.75/night — still the highest weekly averages of the pressure-instrumented window, but pulled down by tonight from the 3-night snapshot (which was 2.52 AHI, 13.0 OA). Two more clean nights would land the week around ~2.05 AHI, which would still be a fresh weekly high. The elevated-week signal is real; tonight moderated it but didn't erase it.
PRIMARY FINDING: CLUSTER BROKE, BUT PATTERN SHIFTED. The 4-night obstructive escalation (1.63 → 2.20 → 2.55 → 2.82) did not extend to a 5th night. AHI fell to 1.54 and OA to 8, both at or below 30-day baseline. This removes the immediate escalation trigger. However, the night wasn't a full return to normal: 3 RERAs appeared (vs. 0 across the entire cluster), and pressure ran ~0.3 cmH₂O higher than 08-21's despite half the discrete events. Read together, the auto-titrator saw an airway-resistance signature (RERAs = upper airway partially closing without meeting hypopnea/OA criteria) and pushed pressure to compensate. That's a different physiological signature than the 08-19/20/21 discrete-obstruction pattern — same underlying "something's off in the airway" story, different presentation. The list of candidate drivers from the 08-21 report (allergen load, congestion, position, weight, alcohol, medication, reflux, airway-tone drift) is unchanged and still worth considering.
REVISED CLINICIAN PLAN. The 08-20/21 report escalation trigger — 5th consecutive rising AHI — is not met, so do not urgently contact the clinician this week. Instead, save the 5-night arc + RERA emergence for the next scheduled check-in. If any of the following occurs, upgrade to immediate escalation: (a) AHI ≥ 2.5 on any single night in the next 7; (b) AHI ≥ 2.0 for 3 consecutive nights; (c) RERA count ≥ 3 on 2 or more nights in the next 7 (that would establish a resistance-driven pattern rather than a one-night event); (d) any night with both AHI ≥ 2.0 AND REM % ≤ 15%.
MASK CHANNEL: FORMALLY CLOSED. Three consecutive nights under the mechanical-issue threshold (0.69, 0.20, 0.74). Leak_90 at floor for the last two nights. Stop monitoring this channel unless leak_avg exceeds 1.0 for 2+ consecutive nights. Remove leak from the "flags" section going forward.
PRESSURE CHANNEL: MILD ELEVATION, WORTH WATCHING. P_avg 8.00 is 5th-highest of matched history and +7% vs. 30-day. P_90 10.72 is above the 08-21 report's ≤ 10.5 threshold. This is the machine's read that some form of airway resistance is present. Watch P_avg over 08-23/24 — if it stays ≥ 7.9 while AHI is at or below 2.0, that's consistent with a resistance-driven (RERA-heavy) pattern; if it drops back to the 7.4–7.6 range, the auto-titrator will have concluded tonight was a one-off.
ARCHITECTURE CHANNEL: REM ANOMALY, REST NORMAL. Deep at baseline, core normal, efficiency at 30-day norm. REM at 17.51% is the standout. Most likely explanation is short TIB (6.41 h vs. 30-day 6.70) — a 30-minute-shorter night at the tail end drops the last REM cycle. Secondary contributor is likely the 3 RERAs fragmenting REM. Not a standalone concern tonight, but if REM stays < 20% for 2+ consecutive nights, revisit — the decoupling story we established through 08-21 rests on REM staying near baseline (23–25%), and a sustained REM drop would be the first architecture-side signal that the AHI cluster or RERA signature is starting to matter for sleep quality.
BEDTIME CHANNEL: SLIPPED. 23:35 bedtime is 65+ min past the 22:15–22:30 target the 08-21 report recommended. The immediate cost: −1.24 h TIB, −1.49 h TST, likely the driver of the REM drop. Reinforce the 22:15–22:30 target for 08-23. Tonight is one data point of slippage; two in a row and it becomes a trend.
What to watch on 08-23. (1) AHI/OA — sustained AHI ≤ 2.0 for 3 nights closes the cluster; any rebound ≥ 2.5 reopens it. (2) RERA count — return to 0–1 means tonight's 3 was a one-off; ≥ 2 again establishes a resistance-driven signature that changes the clinical framing. (3) P_avg — if it stays ≥ 7.9, the algorithm is confirming persistent resistance; if it drops back to 7.4–7.6, tonight's pressure was a transient response. (4) REM % — needs to return to ≥ 20% to confirm tonight's drop was TIB-driven; sustained < 20% would open the architecture-side concern. (5) Bedtime — back to 22:15–22:30 or the TIB/REM story compounds.
2026-08-22T23:35:08) marks the start calendar date, which lines up with the OSCAR "Period" for 2026-08-22. Convention continues to hold.