WP-101 · Prematurity as a Fold Event — Principia Orthogona Vol VI

Bio Domain · Prematurity · dm³ · CRNT · Fold Bifurcation

kernel-verified Lean 4, no sorry documented Primary literature model dm³ applied, not independently verified open Conjecture or open problem closed · neg Obligation attempted, did not clear

Cross-references: Ch A · Autophagy (anchor) · Ch 13 · Thymic Selection · Ch 15 · Cardiac Rhythms · Ch 16 · Hopfield Networks · WP-31C · Calibration Pipeline · WP-30 · The Missing Anchor · Preemie Manual (AXLE)

A preterm birth forces the developing organism across multiple biological threshold events before the attractor basins supporting those transitions are fully formed. This paper maps four such thresholds onto the dm³ operator chain G = U∘F∘K∘C, using the kernel-verified autophagy result as the topological anchor. Two CRNT obligations were attempted and closed negative. The boundary of what is actually verified is precisely drawn.

§1 · The Verified Anchor

What is actually kernel-checked

The starting point is what is proved. From AutophagyDm3_v2.lean (24 theorems, no sorry, no True conclusions, AXLE repo; recounted 2026-09-09):

Kernel-verified · AutophagyDm3_v2.lean kernel-verified

Contact coefficient: c(ρ) = −2ρ < 0 for all ρ > 0 — scalar witness that α∧dα ≠ 0 on X_auto. Not the full manifold statement.
Whitney A₁ fold: V(q) = q³ − 3q has a genuine Morse critical point at q = 1, non-vacuously checked. Algebraic half only — kinase data and Mather finite-determinacy remain open (Obligation B).
Gronwall stability radius: ε₀ = 1/3, positive, less than 1, with basin asymmetry 1/3 < 4/5.
Annular basin: [1/3, 2] is compact and non-empty — Heine-Borel on a closed interval. Named "dm³ basin" only pending a semiflow definition (Obligation C).
Lyapunov exponents: μ_canonical = −3, μ_dm3 = −2, both negative.

Three obligations remain open in the file: full contact non-degeneracy on X_auto (A); Whitney A₁ from mTORC1 kinase data (B); limit cycle via Poincaré–Bendixson (C).

SymPy-verified · WP-31C · separate tool documented CRNT deficiency δ = 2 for the AMPK–mTORC1–ULK1 network established by explicit graph construction (n=9, l=4, s=3). Necessary, not sufficient, for fold behavior. Rate constants did not clear (WP-31C: "Estimate: did not clear"). μ_max ≈ −0.41 s⁻¹ withdrawn in WP-30 — inconsistent time constants, untraceable to literature. This paper does not use that parameter.

§2 · Domain 1 · Pulmonary Surfactant

Surfactant synthesis — phenomenological only

The biological threshold documented

SP-B synthesis is functionally absent before approximately 24–26 weeks gestation and reaches adequate levels around 34–36 weeks. The transition is not gradual: below the gestational threshold, respiratory distress syndrome (RDS) is near-universal in the absence of exogenous surfactant; above it, RDS incidence drops sharply. (Jobe & Ikegami 1987; Whitsett & Weaver 2002, NEJM)

Antenatal corticosteroids accelerate surfactant maturation by acting upstream of the TTF-1 threshold — augmenting the C operator pharmacologically. (Roberts & Dalziel 2006, Cochrane) documented

dm³ mapping model — phenomenological only

C = glucocorticoid/mechanical compression; K = TTF-1 activation threshold; F = lamellar body commitment; U = sustained surfactant secretion. The mapping is a phenomenological analogy only.

[LEAN-NEEDED-1] · Closed, negative result · 2026-09-06 closed · neg CRNT deficiency analysis of the TTF-1/NKX2-1 → SP-B/SP-C network (9 reactions, 5 species + ∅, including degradation sinks and biologically honest autoregulation as de novo transcription) returned δ < 0 under two independent formulations. Adjusting the network to force δ = 2 was considered and rejected — fitting topology to a target value invalidates the proof model.

What δ < 0 actually means. Deficiency is δ = n − ℓ − s, and it is non-negative for every reaction network: the reaction vectors of a linkage class span at most one dimension fewer than the class has complexes, so s ≤ n − ℓ. Degradation sinks and de novo synthesis do not change this. A computed δ < 0 is therefore a defect in the graph construction — miscounted complexes, linkage classes, or stoichiometric rank — and cannot be a property of the network. Both formulations were wrong, in the same direction. corrected · 2026-09-09

Conclusion, on the corrected ground. No valid deficiency was ever obtained for the surfactant network, so CRNT supplies no evidence either way about the contact morphism to X_auto. That is weaker than “not supportable by CRNT topology” and it is the direction to err in. The morphism stays withdrawn and no SurfactantDm3.lean will be written on this basis; what is withdrawn with it is the claim to have shown anything against it. The surfactant threshold itself is real and clinically documented, and untouched by either reading.

§3 · Domain 2 · Cardiac Conduction

Bradycardia of prematurity — shallow attractor basin

Preterm infants show high rates of bradycardia of prematurity. The mechanism is incomplete maturation of parasympathetic cardiac control. documented (Poets et al. 1994, Pediatrics; Fairchild et al. 2016, Frontiers in Physiology)

In dm³ terms: the Lyapunov attractor basin for stable sinus rhythm — the limit cycle Γ of Ch 15 — has not fully formed in the preterm autonomic system. The fold at K* is crossed repeatedly because the basin is shallow. model

Falsifiable in principle: HRV data from preterm infants should show lower Lyapunov stability than term infants of matched corrected age, converging toward term values as corrected age increases. Such data exist (Lake et al. 2002, Am J Physiol) but have not been fitted to dm³ parameters. open

§4 · Domain 3 · Neural Attractor Formation

Myelination and Hopfield basin depth

Myelination proceeds most rapidly between 28 and 40 weeks gestation. In the preterm brain this window occurs ex utero. documented (Counsell et al. 2006, Pediatrics; Ball et al. 2013, Cortex)

In Hopfield network terms (Ch 16): the weight matrix W is sparser and less structured than in a term brain at equivalent developmental stage. Energy wells are shallower, network capacity (~0.138N) is lower. model

Preterm infants at corrected term age are widely reported to show different EEG coherence patterns than term newborns, consistent with shallower attractor basins. This claim is currently unsourced in this paper. It cited Stjerna et al. as NeuroImage 2015; the paper of that title is Stjerna et al., J Vis Exp 2012, and it is a recording protocol, which does not support a finding about coherence. A source that does has not been substituted, so the claim is downgraded rather than left standing on a citation that does not carry it. dm³ parameter fitting is a separate open calibration problem. open

Auditory stimulation in the NICU — including the mother's recorded voice and selected music — has documented effects on preterm neurodevelopment. (Loewy et al. 2013, Pediatrics; Filippa et al. 2019, Acta Paediatrica) documented In attractor terms: consistent auditory input during myelination may contribute to attractor depth. Plausible mechanistic reading, not verified. open

§5 · Domain 4 · Immune Proofreading

Thymic selection — asymmetric thresholds

Preterm infants have reduced naive T cell output, lower thymic index, and altered regulatory T cell proportions. documented (Gibbons et al. 2014, Nat Med; Rechavi et al. 2015, Sci Transl Med)

Ch 13 frames thymic selection as C∘K*∘K** — positive selection at K*, negative selection closing at K**. Literature review (2026-09-06) confirms the two-threshold model is real. documented (Moran 2012, Immunology; Daniels et al. 2006, Nature 444:724–729)

Asymmetry finding — thresholds are not symmetric

K** — negative selection: Very sharply defined. A 1.2-fold difference in TCR-pMHC affinity separates the weakest negative selector from the strongest positive selector (Daniels et al. 2006; PNAS 2009). Sharp threshold — consistent with fold structure. documented

K* — positive selection: Graded, not sharp. Positive selection occurs over a broad range of ligand potency. No sharp threshold separating positively selecting ligands from those that bind too weakly. documented Inconsistent with a Whitney A₁ fold — a fold requires sharp commitment, not a graded response.

[LEAN-NEEDED-2] · Closed, partial result · 2026-09-06 closed · partial Two thresholds are real and documented. K** (negative selection) is sharp and phenomenologically fold-like; K* (positive selection) is graded and the fold analogy does not apply. The full contact morphism claim for the C∘K*∘K** cascade is withdrawn.

A phenomenological fold analogy for K** alone is supportable as a documented sharp threshold. A CRNT or ODE characterization of K** specifically remains possible future work.

Editorial note for Ch 13: The K* fold claim in Ch 13 of Vol VI is overstated and should be revised in a future pass. The K** sharp-threshold reading stands.

§6 · Summary

What holds and what does not

Domain	dm³ mapping	Contact morphism to X_auto	Status
Molecular · Autophagy	Full C∘K∘F∘U; contact form α = dz−ρ²dθ; CRNT δ=2	Anchor — not a claim, the reference object	kernel-verified fold algebra · documented CRNT · rate constants open
Pulmonary · Surfactant	C∘K∘F∘U phenomenological	Withdrawn — CRNT δ < 0	documented threshold · closed · neg morphism
Cardiac · Sinus rhythm	Limit cycle Γ; shallow Lyapunov basin	Not attempted — different topology	documented clinical · model basin · open HRV fitting
Neural · Myelination	Hopfield W matrix; shallower energy wells	Not attempted — different topology	documented EEG · model W-matrix · open fitting
Immunological · K** only	Sharp negative-selection threshold — fold-like	Withdrawn for K; K* phenomenological only	documented asymmetry · closed · partial · Ch 13 K* claim flagged

§7 · The Clinical Instrument

Preemie Manual growth tracker

The Preemie Manual implements the most immediate practical consequence of the fold argument: WHO Child Growth Standards are suppressed before corrected age zero and the Fenton 2013 preterm chart is cited instead. A reference standard built for term-equivalent biology does not apply before the relevant thresholds have been crossed. The tracker is valid for any child from term-equivalent age to 5 years (WHO), and for preterm infants from birth using the Fenton chart (22–50 weeks PMA).

§8 · Open obligations

What remains

[OPEN-1] HRV Lyapunov fitting Fit dm³ Lyapunov parameters to published HRV data from preterm infants (Lake et al. 2002 or equivalent) at matched corrected ages. Cardiac timescales are directly measurable — this may be more tractable than the autophagy rate constant problem that defeated WP-30.

[OPEN-2] Neural attractor calibration Map EEG coherence data (a source for which still has to be identified — see the correction note) to Hopfield parameters: does the coherence deficit correspond to reduced W, or increased noise in the energy landscape? These make different predictions about recovery trajectory.

[OPEN-3] K** CRNT characterization The negative-selection threshold K** is sharp and documented. A CRNT or reduced-ODE characterization of the TCR signaling network at K** specifically remains possible — the continuous nature of TCR affinity is the obstacle, not the sharpness of the threshold.

References

Sources

Autophagy anchor: AXLE/AutophagyDm3_v2.lean · WP-30, WP-31C, WP-31D · totogt.github.io/geometry/book6/wp86-autophagy-calibration-case-study
Jobe AH, Ikegami M (1987). Surfactant for the treatment of respiratory distress syndrome. Am Rev Respir Dis 136(5):1256–75
Whitsett JA, Weaver TE (2002). Hydrophobic surfactant proteins in lung function and disease. NEJM 347:2141–8 · PMID 12501227
Roberts D, Dalziel S (2006). Antenatal corticosteroids for accelerating fetal lung maturation. Cochrane Database Syst Rev 2006(3):CD004454
Poets CF et al. (1994). Effect of supplemental oxygen on apnea in preterm infants. Pediatrics 93(4):592–5
Lake DE et al. (2002). Accuracy of heart rate variability measures. Am J Physiol Heart Circ Physiol 283(1):H230–9
Counsell SJ et al. (2006). Diffusion-based tractography in preterm infants. Pediatrics 117(4):1169–74
Ball G et al. (2013). The influence of preterm birth on the developing thalamocortical connectome. Cortex 49(6):1711–21
Stjerna S et al. (2012). Preterm EEG: a multimodal neurophysiological protocol. J Vis Exp 60:3774 · PMID 22371054 — a recording protocol; see the correction note
Loewy J et al. (2013). The effects of music therapy on vital signs, feeding, and sleep in premature infants. Pediatrics 131(5):902–18
Filippa M et al. (2019). Early maternal voice stimulation effects on the prematurely born infant. Acta Paediatrica 109(1):59–65
Gibbons D et al. (2014). Interleukin-8 (CXCL8) production is a signatory T cell effector function of human newborn infants. Nat Med 20:1206–10 · doi 10.1038/nm.3670
Rechavi E et al. (2015). Timely and spatially regulated maturation of B and T cell repertoire. Sci Transl Med 7(316):316ra195
Moran AE (2012). T-cell receptor affinity in thymic development. Immunology 135(4):261–7
Daniels MA et al. (2006). Thymic selection threshold defined by compartmentalization of Ras/MAPK signalling. Nature 444:724–729
Hogquist KA, Jameson SC (2014). The self-obsession of T cells. Nat Immunol 15:815–822

Corrections · 2026-09-09 · from wp101-verify.py

The anchor count. §1 read 18 theorems. AutophagyDm3_v2.lean carries 24, still with no sorry and no True conclusion. The file grew and the quoted figure did not; it is now recounted by the companion script.

The surfactant withdrawal, §2. The withdrawal stands; its stated reason does not. Deficiency δ = n − ℓ − s is non-negative for every reaction network, so a computed δ < 0 reports a construction error rather than a fact about the network, and two formulations returning it were both wrong. The corrected ground is that no valid deficiency was obtained, so CRNT supplies no evidence either way — weaker than what was published.

Four citations. Whitsett & Weaver is 2002, not 2015 (NEJM 347:2141–8, PMID 12501227). Ball et al. 2013 is in Cortex, not Brain. Gibbons et al. 2014 is Nat Med 20:1206–10 with a different title, not J Allergy Clin Immunol. The Stjerna citation named a journal and year that do not exist for that title; the real paper is J Vis Exp 2012 and is a recording protocol, so the EEG-coherence sentence it supported is downgraded to unsourced rather than re-attached to a paper that does not carry it.

Still open. Fairchild et al. 2016 (Frontiers in Physiology) and a “PNAS 2009” are cited in the text with no reference entry, and the Poets et al. 1994 entry could not be resolved to a paper of that title, journal, volume and pages. Those three are unresolved here rather than guessed at.