ReSTAGE persistence: the Utah quantile describes a proxy, not the first passage; the clock adds no node; the ovarian index stays off

1. Executive scientific abstract

M1-3 does not reopen the isolated jump that M1-2-HG-H3 v2 already won, and it does not restore clause (vi) of M1-1-HG-H1 v1. The question of 1.3 is the second threshold: whether |ΔL|≥7 persistence with floor k≥11 arises from (A) a quantile of SD(L) under Huang Λ plus Ecochard’s published ACF at lags 2–4, without inventing diag(Ω); from (B) an extra-ovarian circamonthly clock that adds no growing-cohort node; or from (C) a residual that requires an ovarian index exogenous to bleeding, the selection day anchored to LH or dimeric inhibin B. Hypothesis generation crowned (A) as the generator in a Utah prefix. Adversarial review withdrew that identity. The experimental design showed why the prefix cannot close it. This report freezes that base for REPORT_EN.

(A) is split and is not closed. In the consecutive Utah prefix (md5 9fbde613a3bbe1799ddaebd2a192ecf8; 50 women, 608 cycles, k 11–15, median age 27 years) the mean of 10-cycle blocks with at least two pairs |ΔL|≥7 is 0.113 versus 0.097 under log L ~ N(3.313, exp(−4.795)) iid. The difference is +0.017, IC90% −0.047 to +0.087, and it includes 0. The ACF of ΔL at lags 2, 3 and 4 is −0.034, +0.025 and −0.054; the three intervals cover 0. That is M1-3-AR-H1 v1: a description of the sliding proxy on this floor, maturity H2 of the contrast already run and H0 of the generator. It is not equivalence, it is not −3, and it is not the marker that STRAW+10 and ReSTAGE publish. That marker is a first passage: the first segment whose length differs by more than 6 days from the previous one and that magnitude is observed again within the next 10 segments (Harlow et al. 2008, criterion [1], PMC2225986, body reread this turn). ReSTAGE estimates the age at that marker by Kaplan–Meier, with the censoring age of whoever did not present it. Coding the end of a short file as a negative is not criterion [1]. The sliding block is not rehabilitated.

The same body clarifies a limit that the design already used and that this report makes precise. Among those with an observed FMP, persistent >6 days is 100% in TREMIN (211/211) and 100% in SWAN (132/132). The drop the authors attribute to the 10-segment window and to SWAN’s shorter follow-up (first 6 years) is the drop in the standard-deviation markers: 85.6% and 84.8% in SWAN versus 99.5% and 98.6% in TREMIN. Short follow-up hides markers that require a reference window. It does not hide, in those women who reach FMP, persistent >6 days. Utah’s problem is different and local: with median k 12 and maximum 15, a window of 10 pairs after the first jump closes only if that jump falls in the first k−11 pairs. With k=11 it never closes. Under the already frozen pair rate q=0.049, the null expectation of closed windows in the 50 women is 2.0 to 4.6. The conditional is declared non-identifying below 30 closed windows. That expectation is a design calculation. This turn did not open the CSV and does not score the first passage.

The serial clause of (A) is not added either. Ecochard et al. 2024 (PMID 38598621) publish P values, not coefficients. In North America, series of at least 7 cycles (3 529 cycles / 314 women), the only order that rejects the null is 3 (ACF P=0.0084, PACF P=0.0030). Lags 2 and 4 do not. Utah is the same deposit the paper already tabulates (3 137 cycles / 562 women with conception=No). An IC90% of ΔL in 50 women from a k≥11 subset does not refute that P. Marquette remains HTTP 403 and Padova remains timed out: the European coefficient of lags 2–4 is still unestimated. Europe is not declared dead from n=50. diag(Ω) is not invented.

(B) remains weakened. Yen and Tsai 1971 (PMID 5552412) leave LH and FSH on a plateau at three weeks after oophorectomy, without a rhythm of ~28 days; Yen et al. 1972 (PMID 5012772) preserve ultradian pulses. That constrains a self-sustained monthly timer without an ovary. Ecochard’s actogram is the cumulative sum of L minus the woman’s mean: it redescribes bleeding and adds no growing-cohort node. The North American actograms of ≥18 cycles are 331 cycles / 12 women. Coexistence of three generations in Gougeon (~85 days, PMID 3558758) makes a memory of about three cycles licit. It is not a PACF. Ecochard himself writes that an hourglass of ~90 days would not produce shorter cycles, and that an FSH variant does not explain why length depends on previous lengths. M1-3-SCOPE-H2 v1 is not crowned.

(C) stays off. It would switch on only if the censored first passage, in a series that crosses CP_σ=42.84 years and can close ≥30 windows, remained separated from the published Λ(age) and the level coefficient at lag 3 did not absorb the difference. The index, if that happened, would still be selection day|LH or dimeric inhibin B. The public inventory of that index with bleeding and k≥11 in −3 is 0. The Utah CSV does not have those columns. D1 stays closed. Mucus peak and luteal phase do not substitute.

The translational M1-3-AR-HT1 v1 stays parked. There is no asset, dose or readiness. Switching off recurrent |ΔL|≥7 on a young calendar would still be manipulating a bleeding classifier, and that classifier is not even identified in the prefix. The sentence «there is no node because the proxy fitted Λ» is withdrawn: the measured object does not observe the ovary. Follicular freeze and do(TGFBR3L) remain illicit. G8 closed. HUMAN_QA_REQUIRED. It is not H5. SD(L), the ACF, selection day and inhibin B are not called a biomarker. Current Lua data are not used. Nothing is prescribed.

Scientific delta of this report (M1-3-REPORT-EN-C1): the English twin freezes, after adversarial review and the design, that (A) is not the conjunction Λ + lags 2–4 as generator of the ReSTAGE first passage: the quantile describes a block proxy in Utah k≥11 (0.113 versus 0.097, IC90% that includes 0) and the first passage is a censored time whose null expectation of closed windows in that prefix is 2.0–4.6, non-identifying below 30, without the CSV having been scored; the ReSTAGE body reread today shows that the non-observation attributed to the 10-segment window is that of the standard-deviation markers in SWAN, while persistent >6 days remains at 100% among observed FMPs. (B) adds no growing-cohort node. (C) stays off. D0-PASSAGE is written and not run. There is no Pharma lever.

2. Scientific question and relevance

The question of 1.3 is not whether a start bit exists. M1-1 mapped the −3→−2 crossing and left the generator open. M1-2 split the isolated jump of the second threshold: Huang Λ at 35 years (α0=3.313, β0=−4.795, SD(L)≈2.5 days) plus consecutive independence calibrate P(|ΔL|≥7)≈0.05 per pair, and Ecochard does not reject lag 1. Persistence within 10 cycles is a different object. M1-2 left it as the governing question of 1.3, with the warning not to treat lags 2–4 as if they required Ω.

To answer it, objects that several stages had soldered together, and that this report does not solder again, have to stay split:

1. Is the proxy «≥2 pairs |ΔL|≥7 in a block of 10 cycles» the STRAW/ReSTAGE first passage? No. The published object is the first jump and a second one within the next 10 segments. The age of the marker is the age of that fulfillment.
2. Is an IC90% that includes 0 on the proxy equivalence to Λ, or is it the generator? No. The interval admits an excess on the order of the rate itself.
3. Is the end of an 11–15 cycle prefix a negative of criterion [1]? No. It is administrative censoring if the window did not close. ReSTAGE uses Kaplan–Meier.
4. Are Ecochard’s lags 2–4 a coefficient, a North American conjunction, or a European P? Table 2 is P. In North America the only published significant order is 3. Europe still has no ρ̂.
5. Are three generations of ~85 days a PACF of L? They are architecture. The map L_{t−3}→L_t is not measured.
6. Do Ecochard’s actogram or title demonstrate an extra-ovarian clock? The actogram is a redescription of L. Yen constrains the month without an ovary.
7. Is CP_σ=42.84 years the within-woman operator? No. Huang proposes the change-points as alternative indicators to ReSTAGE. M1-1-K2.
8. Do the median of 27 years, or one woman aged 40–44, identify −3? No.
9. Does absence of a residual in this proxy close (C) or license saying the ovary is superfluous? No. (C) waits for a residual of the first passage. (vi) stays withdrawn.
10. Is delaying the bit, the FMP, or raising AMH success? No. The coprimary remains months of competent ovulatory cycles. G8 closed.

The relevance is identifiability, not a product. A do on |ΔL| without an index exogenous to bleeding manipulates the classifier. A quantile that fits a young proxy does not say which node, if any, moves onset. Knowing that the prefix does not identify the first passage avoids crowning a generator the file cannot see.

3. Scope, population and life stage

The biological estimand the line pursues is the adult woman in natural reproductive aging, with stage written by bleeding regularity. The stratum in which onset would matter is the late reproductive stage approaching the −3→−2 crossing, and the series in which ReSTAGE dated the marker: calendars of years, entry at 35 in TREMIN, age at the marker by Kaplan–Meier, TREMIN median 41.0 years for persistent >6 days. That support is not in the open deposits this project could measure.

What was measured is a different population. Utah Creighton (CEIBA, CMFS, TTP; 1990–2013): women aged 18 to 40 years, not pregnant at entry, regular menstrual bleeding, without exclusive lactation. On the floor k≥11: 50 women, 608 cycles, median age 27 (20–40), 48 / 1 / 1 below 35, aged 35–39 and aged 40–44. Sex: women. The gap is not sex. It is stage, duration and cohort. Natural family planning with regular bleeding at entry is not the transition in which ReSTAGE calibrated time to FMP. One woman aged 40–44 years does not cross CP_σ=42.84 and does not identify −3.

Inferential unit: the woman, not weighted by k. Cycles, overlapping blocks and Huang segments are not n. p>0.05 is not equivalence. Derry’s decade is not a window of 10 cycles. The eventual census among observed FMPs is not the age curve.

Left out: pregnancy, lactation, mixed PCOS, endometriosis, POI, ovarian cancer, oophorectomy as an onset population, iatrogenic gonadal suppression; current Lua data, PHI, PII and private genomes; prescription; FMP or a symptom as the start; AMH as a census of primordials; follicular freeze and do(TGFBR3L) as levers; LPDF-ARA as first; inventing diag(Ω); requesting the TREMIN microdata; opening D1.

This project does not gavage, does not dose and does not recruit. D0-PASSAGE is specified and was not executed this turn.

4. Background and mechanism map

4.1 Definitions that are not fused

• Onset written by bleeding. STRAW+10, body reread in the adversarial review of 2026-10-23 (PMID 22344196, PMC3340904): persistent difference of 7 days or more in the length of consecutive cycles. Persistence is recurrence within 10 cycles of the first cycle of variable length. The system is applied with no age gate. This turn did not reopen that PDF.
• ReSTAGE criterion [1], PMC2225986 body reread this turn: «the first segment whose length is more than six days different than the previous segment and this magnitude of difference is observed again within 10 segments». In integers, >6 and ≥7 coincide. The reprint «≤7 days» (PMC3580996) is not used.
• Seattle rule. ReSTAGE attributes to Mitchell, Woods and Mariella 2000 the repetition within the next 12 months. ReSTAGE adapted it to 10 segments. This report does not declare that 12 months and 10 segments are the same estimator. Mitchell’s PDF was not opened.
• Age at the marker. Kaplan–Meier with the observed age of the event and, if there is censoring, the censoring age. Censoring: hysterectomy, bilateral oophorectomy, hormonal contraception, hormone therapy or chemotherapy. A gap greater than 2 years censors at the gap. Gaps shorter than 2 years are ignored.
• Block proxy, the one compute.md froze before seeing the result and that is not criterion [1]: in each block of 10 cycles (9 pairs), an indicator of ≥2 pairs with |ΔL|≥7; the woman’s rate = mean of her k−9 blocks; estimand = unweighted mean. With k from 11 to 15 there are 2 to 6 overlapping blocks.
• Frozen Λ. log L ~ N(3.313, exp(−4.795)) iid at 35 years. SD(L)≈2.51 days. No diag(Ω). Age sensitivity, only below the elbow: μ(a)=3.313−0.007(a−35), log σ²(a)=−4.795+0.009(a−35). The slope after CP_σ=42.84 is not re-extracted and does not enter Utah.
• Censored first passage, rule frozen in the design and not applied to the bytes. T1 is the first pair with a jump. There is an event if some pair from T1+1 to T1+10 exists and is a jump. If the window reaches 10 pairs and there is no second jump, the negative closed. If the series ends before that, the woman is censored. It is not a 0.
• Dimeric inhibin B. Heterodimer of preantral and small antral granulosa. A reader of mass. It is not a function of L_t. It is not called a biomarker.
• Selection day|LH. Origin of the dominant’s growth segment, anchored to the surge, not to menstruation. Index of (C) if it does not collapse into L. Absent from these calendars.
• Clock (B). Two statements. B1: a monthly timer that would continue without an ovary. B2: an extra-cohort limit cycle that uses steroid and adds no growing-cohort node. Neither is crowned.
• Fertile function. Months of competent ovulatory cycles. Not the bit. Not the FMP. Not high AMH.

4.2 The chain that would have to be shown, and where it is cut

Architecture (H/R, not an L series)
  class 1 with internal theca → class 8          [Gougeon 1986, ~85 d]
  three generations coexist                 [cross-section]
  map L_(t−3) → L_t                          [ABSENT; it is not the PACF]

Projection onto bleeding (H mid-reproductive, I to −3)
  L ≈ FP + LP
  r(L,FP)=0.829 ≫ r(L,LP)=0.255               [Fehring 2006; algebra]
  an origin memory would be read in L            [if the map existed]

Classifier
  first |ΔL|≥7 and another within 10 segments [ReSTAGE, KM, censoring]
  block proxy in Utah k≥11               [0.113 vs 0.097; it is not the former]
  North American order 3                      [P yes; ρ̂ no]

(B)  actogram = cumsum(L − mean)            [redescribes L]
     month without ovary                           [Yen: plateau, ultradian pulses]

(C)  day|LH or dimeric inhibin B             [column absent]
     only after a residual of the first passage        [D0-PASSAGE not run]
     after Λ(age) and level lag 3  [elbow series: not here]

Today no human link anchors the 85-day pipeline, nor an extra-ovarian clock, nor inhibin B, to the first fulfillment of criterion [1] within woman in −3. The quantile at 35 years describes a regular remainder in a young prefix. It does not explain onset.

5. Evidence method

Unit: the woman. Classes: H human, A animal, V in vitro or ex vivo, C computational, I inferred, R review. A review does not vote causality. Polarity is read against the generator of ReSTAGE persistence, not against FMP, not against −3a and not against the isolated jump.

This turn reread the ReSTAGE body in PMC2225986 and confirmed criterion [1], the censoring, the Kaplan–Meier, the Table 2 proportions among observed FMPs, the TREMIN median of 41.0 years and the sentence that before 40 the marker may be a lifetime pattern. PMC11006216 (Ecochard) returned HTTP 403: the P values, the Utah n, the actogram and the hourglass are inherited from the reread of 2026-10-23 in adversarial-review.md, and are not re-extracted today. Huang, Gougeon, Yen, STRAW+10, Derry, Bortot and Harlow–Zeger are inherited from the verification of 2026-10-22 and from the review of 2026-10-23. The Utah numbers are those of compute.md. The design Monte Carlo (q=0.049, seed 20261024, N=100000 per length) did not open the CSV.

No coefficient is invented. A proxy is not called a biomarker. Current Lua data are not used. Nothing is written in memoria agentes/Lua Labs/reportes/.

6. Evidence map

7. Contradictory evidence and null findings

8. Multiscale mechanistic synthesis

The synthesis of 2026-10-22 wrote that a perturbation of FSH, of antral N or of atresia «propagates to L_{t+2…t+4}» and that this is a PACF of orders 2–4. That sentence was the inferred link that turned architecture into a signature. Adversarial review withdrew it as an identity. This report keeps the correct cut.

Molecular and cellular. Dimeric inhibin B is made by small-antral granulosa and does not rise only because FSH rises (Welt 2001, human in vitro). Groome 1996 measures it as a heterodimer across the human cycle. Robertson 2021 sees it linear with antral count (r=0.70) in one interval, in women of late reproductive age whose reference group is 48.7 years: mass between women, not within-woman variance of the first −2. TGFBR3L is the co-receptor in the mouse gonadotrope (Brûlé 2021); the knockout female has more antrals and a larger litter. Human occupancy in −3 remains absent. There is no target this calendar has shown. CLOCK/BMAL1 in porcine granulosa (Liang Huang 2023, PMID 36921565) is a 24-hour clock. It is a negative control of (B), not a month.

Tissue. Gougeon 1986, human histology: the internal theca of the follicle that ovulates differentiated about 85 days earlier. Recruitment of the next ovulatory follicle, in that tradition, occurs 15–20 days earlier, in follicles of 2–5 mm. Three classes can coexist. That is a cross-section. L_t is written by the final stretch of the follicular phase and the luteal phase. For L_{t−3} to enter L_t a map is required that stores the milieu of three cycles ago and converts it into duration. Coexistence is not that map. The class of that implication is I.

Polarity is not inherited automatically either. Ecochard uses an hourglass «primordial → ovulatory ~90 days» that would only lengthen, and therefore prefers the clock. The compartment is cut wrong: primordial to secondary is another stretch, on the order of 290 days in the same tradition, not the 85 of class 1. Correcting the compartment does not donate the sign. Klein 2002 shows that selection can advance, with the FSH-peak→LH slope indistinguishable (10.7 versus 11.0 days, P=0.74) and a shorter follicular phase. That opens shorter cycles as a possibility. A P=0.74 in one cycle, between women, is not the partial autocorrelation of L. Robertson can delay the inhibin B rise. That does not donate ρ̂ either, and it does not live in −3.

System. Fehring 2006 shows that, in women of about 29 years, L resembles the follicular phase more than the luteal phase. Any memory that moved the follicular phase would be read in L by partition algebra, with the buffer r(FP,LP)=−0.323 intact. That leadership is not persistence, and the mean series of 5.2 cycles does not reach the floor. Henry 2024 and Cole 2009, inherited from M1-2, kill the global identity SD(L)≡SD(FP) in mid-reproductive age and do not donate ρ at 36–44 years. SD(FP) anchored to menstruation stays forbidden as an exogenous index.

Where each rival in the title breaks.

• (A) as the conjunction Λ + lags 2–4. In this young Creighton the quantile clause reaches the proxy and the clause of lags 2–4 of ΔL does not add the window. The level order 3 of the North American pool still has no coefficient. The two clauses are not soldered again.
• (A) as generator of criterion [1]. The proxy is not the first passage, and the first passage has not been calculated. With the lengths already published, the null expectation of closed windows is a few units. A non-rejection the file can barely observe does not identify a generator.
• (B). B1 does not write a month after the ovary is removed. B2 has no observable distinct from the actogram. A ρ̂ that excluded 0 would not distinguish B2 from the projection of lag 3 while an index of the growing cohort does not exist.
• (C). It remains the residual question of onset, not a result of these columns. It opens only after a residual of the first passage that Λ(age) and level lag 3 do not absorb.

A-het, in its place. Lisabeth, Harlow and Qaqish 2004 (PMID 15196619; TREMIN n=997) find that the standard deviation rises 2–6 years before the mean. Ferrell 2005 (PMID 16145311) describes Var_w(L) as U-shaped, with a minimum toward 33–40 years. Huang writes σ stable and then a change at CP_σ, about 3.4 years before the change-point of the mean. That shape can saturate a decade prevalence and an eventual census without a memory of three cycles. It has not yet been compared with the censored first passage on the other side of the elbow, because the series that would cross 42.84 years with enough closed windows is not in this project. Declaring that the elbow is the within-woman operator remains forbidden.

9. Computational layer

decision: use. The question computation could move was whether, with the woman as the unit and without inventing Ω, the recomputed ACF and the frozen Λ specify the window of 10 better than iid pairs. Marquette, the file the decision had chosen (raw census 1 665 cycles / 159 women; 95 with at least 11 rows), again received HTTP 403. Padova again timed out. The calendar that did download is Utah: 106 269 bytes, md5 9fbde613a3bbe1799ddaebd2a192ecf8, identical to the published checksum. Python 3.12.0, NumPy 2.4.4, ACF and PACF in the script (Durbin–Levinson), seed 20261023, bootstrap of women N=4000, within-woman permutation N=2000, Monte Carlo null N=4000.

conception=No reproduces 3 137 cycles / 562 women and includes 32 cycles L>60. The strict prefix —stop before pregnancy, empty length, L>60 or a numbering gap that coincides with the date— retains 2 511 cycles. k≥7 / ≥11 / ≥13 / ≥18 = 141 / 50 / 16 / 0. The maximum is 15. There is no mucus-peak day, luteal phase, LH or inhibin B. (C) has no column.

The computation changed the range of the conjunction and did not change onset. At k≥11 the quantile suffices for the proxy and for the pair; differenced lags 2–4 do not add the window; the negative crude ACF is the −1/n bias of the short series (mean of the iid null ≈ −0.08 at lag 1); the permutation of the window falls on the other side of clustering (0.113 versus 0.178, two-sided P 0.036) and is not promoted to generator. At k≥7 the pair rate does exceed Λ. That excess stays outside the estimand: floor 7 is not turned into the marker after having been seen.

What the computation cannot say, and what the review wrote: compatible with Λ on the proxy is not the first passage. What the design added, without opening the CSV: under q=0.049, P(close) is 0, 0.049, 0.096, 0.140 and 0.182 for k = 11, 12, 13, 14 and 15. The null compound incidence at the end of the series, by Monte Carlo of Bernoulli(q), runs from 0.083 to 0.137 in that range. The two extremes compatible with median 12, maximum 15 and the 16 women with k≥13 give 2.0–4.6 closed windows and 4.7–5.6 events. 1−(1−0.049)^10 = 0.395 is the null recurrence if the window closed. The Wald half-width at 90% of a proportion 0.395 is 0.254 with 10 windows, 0.147 with 30 and 0.114 with 50. Below 30 the conditional does not identify. F̂, the length-matched cumulative incidence, is identifying for this short prefix only if the half-width of its IC90% is less than 0.10.

The raw Marquette census (95 / 56 / 12 women with ≥11 / ≥13 / ≥18 rows, maximum 45) has a null expectation of closed windows between 7.8 and 23.2 depending on how each stretch is pushed. They are not 95 windows, it is not the series already cut, and it is not a ρ̂. The 403 still stands. BioNeMo and GEO do not observe |ΔL|.

10. Primary hypothesis

M1-3-AR-H1 v1 — in the Utah k≥11 prefix the proxy is compatible with Λ; that is not the first passage

Role. Revised primary of M1-3-HG-H1 v1. It keeps the non-rejection of the frozen margins. It loses the word generator. It is the reading that survives of branch (A) after the conjunction is split.

Statement. In the consecutive prefix already measured (md5 9fbde613a3bbe1799ddaebd2a192ecf8, 50 women, 608 cycles, median age 27 years): (i) the difference between the mean of blocks with ≥2 pairs |ΔL|≥7 and the iid log L ~ N(3.313, exp(−4.795)) has an IC90% that includes 0 (0.113 versus 0.097; difference +0.017; IC90% −0.047 to +0.087); (ii) the ACF of ΔL at lags 2, 3 and 4 has IC90% intervals that include 0; (iii) (i) and (ii) do not identify the first passage of STRAW+10 / ReSTAGE, do not equate to Λ, do not refute the North American order-3 P, do not restore (vi) and do not name −3.

Mechanism. Statistical description of a regular remainder. Λ is the null of the proxy in young women with regular bleeding at entry and series of 11 to 15 cycles. It is not the cause of the marker that ReSTAGE dates toward 41 years in TREMIN.

Predictions. The same proxy, in Marquette or in Padova, may repeat the non-rejection or stop doing so. Either of the two results is still not the first passage. The first passage is judged in D0-PASSAGE, not by rereading these intervals. If D0-PASSAGE becomes identifying and the IC90% includes 0 on the floors k≥7 and k≥11, AR-H1 remains a local description of that incidence in this young Creighton, still without being the age at the marker and without being −3.

For. The numbers in compute.md. Class C, public human calendar, direct for the proxy on this floor. The within-woman SD covers the frozen SD(L). The isolated pair fits the mean-field. Class H: the published criterion is a different object (ReSTAGE, reread this turn; STRAW+10, inherited the same day).

Against. The IC90% is wide. The floor k≥11 thins the tail. The file was already inside Ecochard. The stage is not the transition. The permutation of the window (P=0.036) prevents reading the order as pure iid and is not used to rescue the hypothesis.

Kill. Declare equivalence or −3 from this IC90%. Move the estimand to floor 7 after seeing the excess of pairs. Treat (i) as the first passage. Code censoring as 0 and call that a confirmation.

Status. weakened. Maturity. H2 of the rate contrast already run. H0 of the generator and of the first passage. Ordinal confidence. 0.70 that the frozen margins of the proxy are not met on this floor. 0.15 that this identifies the ReSTAGE marker. It is not H5.

What remains of (A) as an answer to the question. The heteroscedastic quantile accounts for the proxy in this file. Ecochard’s published ACF at lags 2–4 is not added here, and in North America the paper did not publish that conjunction: it published order 3, without a coefficient. No Ω. No generator of the second threshold.

11. Competing hypothesis

M1-3-AR-HSEL v1 — the fit to Λ is the survival filter of the prefix

Role. Competitor of the reading «Λ suffices». It is not an ovarian mechanism. It explains the result of the run and it is rank 2 after the attack, ahead of the clock.

Statement. Compatibility of the proxy and of the pair rate with Λ at k≥11 depends on having required eleven consecutive cycles after cutting at pregnancy, a gap or L>60, inside a cohort with regular bleeding at entry. On the floor k≥7 the pair rate separates from Λ upward (0.088 versus 0.049, IC90% 0.020 to 0.058). The marker, defined on women in whom the recurrent jump occurs, lives in the tail that long floor thins.

Mechanism. Selection. Entry asks for regularity. The series stops before pregnancy or a cycle >60 days. No prefix cut by L>60 reaches k≥7. Inside k≥7, pregnancy (0.055, n=22) and gap (0.062, n=20) fit Λ; the excess is in whoever reaches the end of the record (0.103, n=97). Asking for k≥11 returns the pair rate to 0.052. A narrow quantile has to win where the marker almost does not visit.

Predictions. The first passage, calculated with censoring and without requiring the series to reach 11 cycles, may show an incidence greater than the iid of Λ in the same Utah, concentrated in the series that end with the record. If the first passage at k≥7 and at k≥11 falls inside Λ with an identifying contrast, this competitor falls and AR-H1 is strengthened as a description of that incidence, still without being −3. If k≥7 excludes 0 and k≥11 does not, the quantile stops traveling as an answer to the marker.

For. Class C, the pair contrast already persisted. Class H, entry criteria in Ecochard 2024.

Against. Floor 11 was frozen before seeing the result; it is not moved today. The excess at floor 7 may be another tail and not criterion [1]. The prediction about the first passage is H0 until D0-PASSAGE runs. Reaching 30 closed windows would already, by itself, be an excess of early first jumps relative to q=0.049. The rule does not assume that result.

Kill. First passage compatible with Λ on both floors, with an identifying half-width. Or move the estimand to floor 7 today and call that a confirmation.

Status. proposed. Maturity. H2 of the pair contrast by floor. H0 of the first passage. Ordinal confidence. 0.55 that the filter contributes to the fit. 0.35 that the first passage will confirm it. It is not H5.

Rivals the question names that do not occupy rank 2

(B) / M1-3-SCOPE-H2 v1, weakened. An extra-ovarian circamonthly clock that adds no growing-cohort node is not the live competitor. B1 is constrained by Yen: the axis that survives without an ovary does not write a month. B2 wins no observable distinct from the actogram, and the actogram does not add the growing cohort because it is a function of L. Lunar phase is not an endpoint; the internal contradiction new moon / full moon is kept (M1-3-EV-C5). Residual ordinal confidence: 0.10 for B2 on bleeding calendars of the Utah type. It receives no arm in the design. A coefficient of lag 3, if it someday excludes 0, would not distinguish it from AR-HEU while the cohort index is not measured.

M1-3-AR-HEU v1, narrow biological competitor, weakened. In the North American pool the level ACF/PACF coefficient at lag 3, with the woman as the unit, might have an IC90% that excludes 0. Even if it did, attributing it to Gougeon requires a map that neither the paper nor Utah has measured. Padova remains the only place where the European conjunction 2–4 could exist as a number. A timeout is not a coefficient. Status: weakened. Maturity: H1 of the published P; H0 of the coefficient and of the map. Ordinal confidence: 0.40 that the level coefficient at lag 3 in the pool excludes 0; 0.20 that, if it does, the reading is Gougeon. Kill: that coefficient covers 0 after the short-series permutation. An IC90% of ΔL in 50 women is not that kill. Utah’s lag 3 is already spent: the level ACF does not separate from the permutation and the ΔL ACF covers 0.

M1-3-AR-HCP v1, schedule null, not a mechanism. If a public calendar appears that crosses 42.84 years and can close ≥30 windows, the first passage against the published Λ(age) —post-elbow slope copied then from the PDF, without refitting and without Ω— is the contrast this hypothesis has to lose or win. Utah does not vote. Declaring that CP_σ is the within-woman operator kills it by category even if the numbers fit. Status: weakened as a mechanism; retained as a null. Maturity: H1 of the shape of Λ; H0 of the window on the other side of the elbow. Ordinal confidence: 0.25 that the post-elbow slope is necessary for the first passage; 0.15 that it is sufficient. The inherited PPD 0.28 threatens the clause «no lags».

12. Translational hypothesis

M1-3-AR-HT1 v1 — a target is not judged by switching off the young bit; the growing-cohort lever is not designed while (C) is off

Role. Parked translational. It inherits the veto of M1-3-HG-HT1 v1 and withdraws the sentence «there is no node».

Statement. An experimental target is not judged by switching off the young proxy or by switching off the first passage on a reproductive-age calendar. Follicular freeze and do(TGFBR3L) remain forbidden. The only research territory that would exist, if M1-3-AR-H3 v1 switched on and the index were not collinear with L, would be scored in months of competent ovulatory cycles and in the dispersion of selection day|LH, not in the bit, not in the FMP and not in an AMH census. While (C) is off that lever is not designed. Compatibility of the proxy with Λ does not demonstrate that the ovary is superfluous.

(C) / M1-3-AR-H3 v1, specified and off. After D0-PASSAGE, if the first passage remains separated from Λ(age) and level lag 3 does not absorb it, selection day anchored to LH or dimeric inhibin B lead that residual and a function of L_t does not. Today there is no first-passage residual, no column and no inventory. Status: proposed, not activated. Maturity: H1 of inherited architecture (M1-2-HG-H1 v2, origin jitter without a buffer-failure trigger); H0 of the anchor. Ordinal confidence: 0.20 that the first-passage residual exists; 0.14, inherited, that the index leads it. HUMAN_QA_REQUIRED before any new woman. It is not H5. It is not a biomarker. D1 — six cycles, ultrasound, inhibin B — stays closed: six cycles do not estimate this marker, and eleven or fifteen in this prefix do not close the window in expectation either.

Status of HT1. parked. Maturity. H1 of the inherited vetoes. H0 of the lever. Ordinal confidence. 0.75 in the veto on dosing the young bit. 0.20 that the lever will come to be written. HUMAN_QA_REQUIRED. It is not H5.

No compound, dose, schedule or negotiation is proposed. No feature, persona or piece of content is proposed. Bone, brain, endometrium and longevity are not fused with this title. G8 stays closed: delaying FMP is not success, and Ruth et al. 2021 (PMID 34349265) already allocates bone and less diabetes against hormone-dependent cancer at the natural age of menopause.

13. Falsifiable predictions and kill criteria

Criteria that kill the report if someone crosses them while reading it: treat 2.0–4.6 as the observed count from the CSV; cite 0.088 versus 0.049 as confirmation of AR-HSEL; declare Europe dead or order 3 dead from n=50; crown P=0.036 or lag 5; read a wide interval that includes 0 as equivalence; identify −3 with the median of 27 years; invent the post-elbow slope; restore (vi); mark H5; call the result a biomarker.

14. Discriminating experiment

The minimal sequence is already written. This report does not run it.

D0-PASSAGE     Utah, same bytes, censoring rule
               not run
       │
       ├─ n_cerradas < 30          the conditional does not identify
       ├─ half-width of F̂ ≥ 0.10   does not identify either
       ├─ half-width < 0.10 and the IC90% includes 0 on the two floors
       │     AR-HSEL falls for this short functional
       └─ half-width < 0.10 and k≥7 excludes 0 while k≥11 does not
             AR-HSEL is strengthened

D0-COEFF       Marquette the day the CSV leaves 403
               Padova the day the host answers
               NA is not mixed with EU
               Utah lag 3 is not looked at again to confirm it

D0-CP          only a public series that crosses 42.84 years
               and can close ≥30 windows
               post-elbow slope COPIED from the PDF, without Ω

D0-INVENTORY   only if D0-CP leaves a residual and lag 3 does not absorb it
D1             closed
Clock          no arm
HT1 lever      no protocol

Population of D0-PASSAGE. The prefix already hashed. It is not a STRAW stage. Replicates, not this turn: Marquette and Padova, after recomputing the hash, in separate analyses. Ecochard’s North American pool is not declared reproduced with the 50 from Utah.

Null model. The same observed k. Λ at 35 years, without Ω. Age sensitivity uses the coefficients already used in compute.md and does not enter the elbow. The within-woman permutation is not the primary control of this functional: it preserves the multiset and answers another question. The already spent proxy permutation is not recycled.

Assay. There is no laboratory assay. Whoever runs the script does not add an LH column or an inhibin column.

Strata, written before the count. Floor k≥7 (141 women in the run already done) and floor k≥11, side by side. Stop reason: end of record, pregnancy, gap. Age <35 / 35–39 / 40–44; if the stratum has fewer than 30 women there is no IC90%. The 40–44 cell of floor 11 is one woman.

Censoring table, mandatory. No T1; event with closed window; closed negative; event inside a window that did not reach 10; censoring with open window. A run without that table is invalid.

Primary estimand, F̂(H). Kaplan–Meier cumulative incidence of the compound event on the pair scale, censored at k−1. H=10 and H=14 are reported. Unit: the woman. Interval: percentiles of the contrast against the null, not a Wald. Bootstrap and Monte Carlo null N=4000. Seed of the future run, declared and not spent on the CSV: 20261024.

Conditional estimand. Only among closed windows, against 0.395. It is not calculated by throwing the censored into the denominator: an early second jump in a short series would be seen and the corresponding negative would not, and that inflates the event.

What is not an endpoint. The sliding block. Derry’s decade. The actogram. Lunar phase. The mucus peak. The luteal phase. Selection day. Inhibin B. Lag 5. P=0.036.

Failures that invalidate the call. Code censoring as 0. Use the block and call it the first passage. Cite the excess of pairs as if it were this functional. Read a wide IC90% that includes 0 as equivalence or as generator. Identify −3 with the median of 27 years. Invent Ω. Activate (C), open D1, request TREMIN or use Lua data because the conditional did not identify. The G-QA gate stops the work: if a new woman were required, it stops. HUMAN_QA_REQUIRED.

Why this is the smallest discriminant. Any «compatible with Λ» on 11–15 cycles may be the quantile or may be that the window has not had time to close. Splitting that ambiguity does not require recruiting. It requires not treating the end of the record as a negative. If the contrast does not identify, the honest result is non-identifying, not a null of the ovary and not a generator.

15. Biomarkers and stratification

Nothing measured in M1-3 is called a biomarker.

SD(L) under Λ is a population parameter of a length model, copied from Huang and used as a null. It is not an anchor of criterion [1] within woman. The ACF is a coefficient, or a P when the coefficient was not published. It is not a reader of onset. Selection day|LH and dimeric inhibin B are the indices (C) would name if it switched on. Today they are not observed together with bleeding at the floor and the stage that would be required. Calling them a biomarker would advance a validation that does not exist: there is no context of use, no repeatability, no gain over bleeding itself.

Strata that do matter, and that are not fused:

• Floor k≥7 versus k≥11. The first is that of Ecochard’s North American ACF. The second is that of the window of 10. The fit to Λ appears in the second and not in the pair rate of the first.
• Stop reason. The excess of pairs at floor 7 lives at the end of the record, not in pregnancy or in the gap, and no cut by L>60 reaches the floor.
• Age. <35 is almost all of floor 11. 35–39 and 40–44 are one woman each. There is no stratum IC90%. They do not identify −3.
• Continent. NA and EU are not averaged. The North American order 3 and the European conjunction 2–4 are different questions.
• Pre-elbow and post-elbow. Utah lies entirely below 42.84 years. ReSTAGE’s pre-40 clause is prognosis to FMP, not a stratum that turns the quantile at 27 years into the marker.

Visit FSH, in ReSTAGE, is associated with all the proposed bleeding markers, with the highest odds for persistent >6 days, and remains associated after age. That does not identify the node: the authors say the early «may be best described by ovarian activity» and they do not measure ultrasound or inhibin. Burger, inherited from M1-1, denies a visit endocrine marker that separates early from late. The two sentences are kept. Neither opens (C).

16. Individual variability

The unit is the woman, and the women in this file are not interchangeable with TREMIN.

Utah enters with regular bleeding and leaves the estimand as soon as the cycle lengthens too far, there is a pregnancy or a cycle is missing. The within-woman SD at k≥11 has mean 2.34 days and median 1.94: the median is narrower than the SD(L) at 35 years, as expected of someone enrolled for regularity and still far from the elbow. At k≥7 the mean SD rises to 2.76 and the median stays at 2.07: the tail pulls the mean. Conditioning on eleven consecutive cycles keeps the center of that distribution.

Ecochard writes that the correlation is sometimes positive and sometimes negative, and that on average it differs from zero. Short North American series, he says, give less precise estimates. Heterogeneous sign plus n=50 is the regime in which an interval covers 0 even if the pool P does not. Level lag 5 separates from the permutation with a coefficient of +0.021 whose interval against 0 includes 0: heterogeneity of order, not a finding.

Harlow–Zeger, in college women aged 17–19 years and segments of 17–43 days, reads a flat autocorrelation as a random intercept (p≈0.3), not as an AR of orders 2–4. Bortot, on the English series, finds that most women are nearly stationary and a minority are not, and that after removing the walk the higher orders become negligible. Huang, after each woman’s own mean and σ(t), leaves a residual PPD of 0.28 with no dominant lag. Three readings of heterogeneity threaten transport of a crude European PACF to onset. None is turned, in this report, into a new ρ̂.

Age at entry in Ecochard is mostly under 35 years (Europe ~83%; North America, 40–50 years at entry <1%, 4 women). Derry shows that the STRAW operator is already met, by decade, in the 20s and in the 30s. ReSTAGE shows that meeting it before 40 does not bring FMP closer. The variability the bit reads in youth and the variability that accelerates after the elbow may be the same bleeding on two schedules. This project does not have the series that would separate them. It does not fuse them.

17. Pharma relevance and maturity

There is no asset a laboratory could license from this report. The object ReSTAGE uses as early has not been estimated on the only calendar the project could measure, and the support where it was estimated —follow-up of years, age at the marker by Kaplan–Meier— is not in a public file usable here. Switching off recurrent |ΔL|≥7 on a young calendar would still be manipulating a classifier. The design reinforces the veto: the published classifier is not identifiable in the prefix, and the expectation of closed windows stays below the floor of 30.

There is no context of use, clinical utility, net benefit, external replication or partnering readiness. It is not H5. HUMAN_QA_REQUIRED. The maturity of the project, taken together, is H2 for the proxy contrast in Utah k≥11, H1 for the published P values and for the vetoes, and H0 for the first passage, for the lag-3 coefficient, for the clock as a title and for any lever.

18. Limitations

• The first passage is still not calculated on the bytes. The expectation of 2.0–4.6 closed windows is the null under q=0.049, not the observed count. A real excess of early jumps could reach 30 windows. The rule already says what to do if that happens. The extremes use the median, the range and the count k≥13, not the woman-by-woman histogram. The Monte Carlo draws Bernoulli(q), not log-normal lengths. The rate 0.049 already summarizes the log-normal null of the pair.
• «Within 10 segments» was frozen as the next 10 pairs after the first. The sentence admits an inclusive-window reading, one segment shorter. Changing it after seeing the jumps would invalidate the call. A sensitivity to 9 pairs may be reported alongside and does not replace the primary.
• This turn reread ReSTAGE and did not reread Ecochard: PMC11006216 returned 403. If the reread of 2026-10-23 had failed on a digit of Table 2, the identity attack does not depend on that digit: it depends on Table 2 being a table of P, a fact already audited, and on the wording of criterion [1], confirmed today. Gougeon, Huang and STRAW+10 were not reopened today. The post-elbow slope is still not copied.
• In the PMC HTML, the SD >8 day row lists 211 and 98.6% under a label the converter wrote «TREMIN TRUST». The other rows of that column are TREMIN. This report reads it as TREMIN, not as a fifth cohort.
• n=50, k 11–15, overlapping blocks, wide IC90%. «Does not exclude 0» is not equivalence. Marquette 403 and Padova timed out are not null coefficients.
• Natural family-planning cohort, enriched in the region where |ΔL|≥7 is rare. That Λ describes the proxy here does not say that it describes the first passage after the variance change.
• The SWAN attribution (85.6% and 84.8%) holds for the standard-deviation markers. Persistent >6 days is 100% in the 132 SWAN women with observed FMP. Using that drop as if criterion [1] itself were lost in a follow-up of 6 years among those who reach FMP would be another object error. The censoring lesson stays in what the article says, and Utah’s non-identification stays in the arithmetic of k≤15.
• Laboratory memory is not consolidated. Nothing is written in memoria agentes/Lua Labs/reportes/.

Nothing is prescribed. Current Lua data are not used. A calendar of fifteen cycles is not the follow-up in which ReSTAGE dated the marker, and the lack of closed windows does not demonstrate that the ovary is superfluous.

19. Conclusions

The question was whether |ΔL|≥7 persistence (k≥11) arises from (A), from (B) or from (C). The answer the verified artifacts allow to be frozen is this:

1. (A) is not, in this file, the conjunction M1-2 left open. Huang’s quantile at 35 years describes the block proxy and the isolated pair in the Utah prefix k≥11. Lags 2–4 of the ACF of ΔL do not add that window. The only North American serial P Ecochard published is order 3, without a coefficient, in a pool this subset does not refute. Europe still has no ρ̂. Ω is not invented. And the proxy is not the first passage. M1-3-AR-H1 v1 is the primary, weakened: H2 of the contrast, H0 of the generator.
2. The first passage is a censored time, and this prefix does not identify it in expectation. ReSTAGE estimates it by Kaplan–Meier. With q=0.049 the null expectation of closed windows is 2.0–4.6. Below 30 the conditional does not identify, and F̂ does not identify if the half-width of the IC90% is ≥0.10. D0-PASSAGE is written and not run. Treating 2.0–4.6 as a CSV result would be false. The competitor of that fit is M1-3-AR-HSEL v1: the long floor keeps the regular remainder. It has not yet been tested on criterion [1].
3. (B) adds no growing-cohort node. The actogram redescribes L. Yen constrains the month without an ovary. Gougeon names a pipeline of ~85 days and does not identify it with a PACF. The clock remains weakened, with no arm.
4. (C) stays off. Selection day|LH or dimeric inhibin B would enter only after a residual of the first passage that Λ(age) and level lag 3 do not absorb, in a series that crosses the elbow and can close ≥30 windows. Inventory = 0. D1 closed. M1-3-AR-HCP v1 waits for that series as a schedule null, not as a mechanism, and does not replace the operator.
5. There is no lever. M1-3-AR-HT1 v1 parked. Freeze and TGFBR3L illicit. G8 closed. (vi) withdrawn. It is not H5. It is not a biomarker. It is not a dose. HUMAN_QA_REQUIRED before any new woman.

M1-2-HG-H3 v2 remains owner of the isolated jump. The clause that required lags 2–4 for all persistence is not restored, and it is not declared dead in the North American pool either. Onset, where σ changes after 42.84 years, is not in the sample of 27 years. The smallest next step is not to recruit or to dose. It is D0-PASSAGE on the bytes already measured, with the censoring written before counting.

20. References

1. Harlow SD, Crawford S, Dennerstein L, Burger HG, Mitchell ES, Sowers MF. The ReSTAGE Collaboration: defining optimal bleeding criteria for onset of early menopausal transition. Fertil Steril. 2008. PMID 17681300. PMC2225986. DOI 10.1016/j.fertnstert.2007.02.015. Body reread this turn: criterion [1]; Kaplan–Meier; censoring; Table 2 among observed FMPs; TREMIN median 41.0 years; pre-40 clause; attribution of the ~85% of the SD markers in SWAN to the window of 10 segments and to the shorter follow-up.
2. Harlow SD, Gass M, Hall JE, et al. Executive summary of the Stages of Reproductive Aging Workshop + 10: addressing the unfinished agenda of staging reproductive aging. J Clin Endocrinol Metab. 2012;97:1159–1168. PMID 22344196. DOI 10.1210/jc.2011-3362. PMC3340904. Body reread in ADVERSARIAL_REVIEW on 2026-10-23. Not reopened today. The reprint PMC3580996 («≤7 days») is not used.
3. Ecochard R, Stanford JB, Fehring RJ, Schneider M, Najmabadi S, Gronfier C. Evidence that the woman's ovarian cycle is driven by an internal circamonthly timing system. Sci Adv. 2024;10:eadg9646. PMID 38598621. PMC11006216. DOI 10.1126/sciadv.adg9646. Inherited from the reread of 2026-10-23: this turn PMC returned HTTP 403. Table 2 = P. NA order 3, ACF P=0.0084, PACF P=0.0030, series ≥7, 3 529 cycles / 314 women. Utah 3 137 / 562. NA actograms ≥18 cycles: 331 / 12.
4. Huang X, Elliott MR, Harlow SD. Modeling menstrual cycle length and variability at the approach of menopause by hierarchical change point models. J R Stat Soc Ser C Appl Stat. 2014. PMID 24729638. PMC3979630. DOI 10.1111/rssc.12044. α0=3.313, β0=−4.795, CP_σ=42.84. Post-elbow slope not re-extracted. Inherited.
5. Mitchell ES, Woods NF, Mariella A. Three stages of the menopausal transition from the Seattle Midlife Women's Health Study. Menopause. 2000;7:334–349. DOI 10.1097/00042192-200007050-00008. Cited by ReSTAGE as reference 4: repetition within the next 12 months. PDF not opened.
6. Gougeon A. Dynamics of follicular growth in the human. Hum Reprod. 1986;1:81–87. PMID 3558758. DOI 10.1093/oxfordjournals.humrep.a136365. Inherited from the abstract already read on 2026-10-22. Architecture of ~85 days. It is not the map of lag 3.
7. Yen SSC, Tsai CC. The effect of ovariectomy on gonadotropin release. J Clin Invest. 1971;50:1149–1153. PMID 5552412. Inherited. Plateau at 3 weeks. n=10.
8. Yen SSC, Tsai CC, Naftolin F, Vandenberg G, Ajabor L. Pulsatile patterns of gonadotropin release in subjects with and without ovarian function. J Clin Endocrinol Metab. 1972;34:671–678. PMID 5012772. DOI 10.1210/jcem-34-4-671. Inherited. Ultradian pulses.
9. Najmabadi S, Schliep KC, Simonsen SE, Porucznik CA, Egger MJ, Stanford JB. 2020. PMID 32104920. Origin of the Utah set, via Ecochard. The title is not reopened this turn. Deposit DOI: 10.7278/S50d-4gxs-s4hj. md5 9fbde613a3bbe1799ddaebd2a192ecf8.
10. Derry PS, Derry GN. 2014. Women's Reprod Health 1(1). DOI 10.1080/23293691.2014.901797. TREMIN n=144. 87% / 78–79% / 100% by decade. Inherited from the verification of 2026-10-22. No PMID in this project's record. The title is not reopened this turn.
11. Harlow SD, Zeger SL. 1991. J Clin Epidemiol 44:1015–1025. PMID 1940994. DOI 10.1016/0895-4356(91)90003-r. ACF 0.21–0.36 on segments of 17–43 d. Inherited from the verification. The title is not reopened this turn.
12. Bortot P, Masarotto G, Scarpa B. Sequential predictions of cycle lengths. Biostatistics. 2010;11:741–755. PMID 20400622. DOI 10.1093/biostatistics/kxq020. α=−0.61 after detrend. Inherited.
13. Klein NA, Harper AJ, Houmard BS, Sluss PM, Soules MR. Is the short follicular phase in older women merely a consequence of advanced recruitment? J Clin Endocrinol Metab. 2002. PMID 12466381. 10.7 vs 11.0 d, P=0.74. Inherited.
14. Robertson DM, Lee CH, Baerwald A. 2021. FSH, inhibin B, AFC and AMH in late reproductive age. Endocrinol Diabetes Metab 4:e00172. PMID 33855196. PMC8029535. LRG 48.7±1.37 y. Inherited from M1-2 and from the synthesis of M1-3. It is not an anchor of −3.
15. Fehring RJ, Schneider M, Raviele K. Variability in the phases of the menstrual cycle. J Obstet Gynecol Neonatal Nurs. 2006. PMID 16700687. r(L,FP)=0.829. mean k 5.2. Inherited.
16. Lisabeth LD, Harlow SD, Qaqish B. A new statistical approach demonstrated menstrual patterns during the menopausal transition did not vary by age at menopause. J Clin Epidemiol. 2004;57:484–496. PMID 15196619. DOI 10.1016/j.jclinepi.2003.10.005. σ rises 2–6 years before the mean. TREMIN n=997. Inherited.
17. Ferrell RJ, O'Connor KA, Rodríguez G, et al. Monitoring reproductive aging in a 5-year prospective study. Menopause. 2005;12:567–577. PMID 16145311. Var_w(L) U-shaped, minimum 33–40 y. Inherited from M1-2.
18. Welt CK, Smith ZA, Pauler DK, Hall JE. Differential regulation of inhibin A and inhibin B. J Clin Endocrinol Metab. 2001. PMID 11232020. Inherited.
19. Groome NP, Illingworth PJ, O'Brien M, et al. Measurement of dimeric inhibin B throughout the human menstrual cycle. J Clin Endocrinol Metab. 1996. PMID 8636341. Inherited.
20. Brûlé E, Wang Y, Li Y, et al. TGFBR3L is an inhibin B co-receptor that regulates female fertility. Sci Adv. 2021. PMID 34910520. Inherited. Litter sign contrary to a delay lever.
21. Huang L, Zhang L, Shi S, et al. Mitochondrial function and E2 synthesis are impaired following alteration of CLOCK gene expression in porcine ovarian granulosa cells. Theriogenology. 2023;202:51–60. PMID 36921565. Porcine granulosa, 24 h. Negative control of (B). It is not Xiaobi Huang 2014.
22. Abel TW, Rance NE. Stereologic study of the hypothalamic infundibular nucleus in young and older women. J Comp Neurol. 2000;424:679–688. PMID 11050378. DOI 10.1002/1096-9861(20000904)424:4<679::AID-CNE9>3.0.CO;2-L. Inherited from the synthesis. +1, not a clock of −3.
23. Rometo AM, Krajewski SJ, Voytko ML, Rance NE. Hypertrophy and increased kisspeptin gene expression in the infundibular nucleus of postmenopausal women and ovariectomized monkeys. J Clin Endocrinol Metab. 2007. PMID 17488799. Inherited from the synthesis.
24. Ruth KS, Day FR, Hussain J, et al. Genetic insights into biological mechanisms governing human ovarian ageing. Nature. 2021. PMID 34349265. G8. Inherited.
25. Weenen C, Laven JS, Von Bergh AR, et al. Anti-Müllerian hormone expression pattern in the human ovary: lack of expression in primordial follicles. Mol Hum Reprod. 2004. PMID 14742691. AMH does not census primordials. Inherited from M1-1.
26. Harlow SD, Lin X, Ho MJ. Analysis of menstrual diary data across the reproductive life span. J Clin Epidemiol. 2000. PMID 10941950. Inherited. Variance context, not Ecochard’s ACF.

Claims of this report: M1-3-REPORT-EN-C1 (delta: (A) split, the first passage is a censored time with null expectation 2.0–4.6 and not run; (B) weakened; (C) off; no lever); M1-3-REPORT-EN-C2 (ReSTAGE PMC2225986 reread: criterion [1], Kaplan–Meier, censoring; persistent >6 days is 100% in TREMIN 211 and SWAN 132; the drop attributed to the window of 10 segments is that of the SD markers, 85.6% and 84.8%); M1-3-REPORT-EN-C3 (the proxy 0.113 versus 0.097 is not the generator or equivalence; lags 2–4 of ΔL cover 0; the North American order 3 still has no coefficient; Europe is not declared dead; Ecochard’s PMC at 403 this turn); M1-3-REPORT-EN-C4 ((B) adds no node: actogram = cumsum of L; Yen constrains B1; Gougeon is not the identity of the PACF); M1-3-REPORT-EN-C5 (D0-PASSAGE not run; conditional non-identifying if n_cerradas < 30; F̂ non-identifying if the half-width ≥ 0.10; 2.0–4.6 is not the CSV count); M1-3-REPORT-EN-C6 (Pharma with no asset; AR-HT1 parked; (C) off; D1 closed; (vi) withdrawn; it is not H5; it is not a biomarker; HUMAN_QA_REQUIRED).

Hypotheses touched: M1-3-AR-H1 v1 (weakened primary); M1-3-AR-HSEL v1 (competitor); M1-3-AR-HEU v1 (narrow biological rival); M1-3-AR-HCP v1 (schedule null); M1-3-AR-H3 v1 ((C), off); M1-3-SCOPE-H2 v1 ((B), weakened); M1-3-AR-HT1 v1 (translational parked, HUMAN_QA_REQUIRED). Lineage that is not rewritten: M1-3-HG-H1 v1 split; M1-2-HG-H3 v2 owner of the isolated jump; (vi) withdrawn.

Nothing is written in memoria agentes/Lua Labs/reportes/. Laboratory memory is not consolidated. Nothing is prescribed. Current Lua data are not used. Next stage: BILINGUAL_QA.