Joint likelihood for correlated observations O-103, O-104, O-105, O-106, O-107 (shared basis: S-47). Step 8 writes the single ## Likelihood block here; the member edges point back via group.
Likelihood
# CG-28 (HC-9) — ONE joint estimate over E-46+E-47+E-48+E-53+E-54: all five are the same Miller 2014 n=6
# dose-response (shared basis S-47), priced jointly (rule 1). Whole pattern: egg ingestion raises plasma/urine
# TMAO dose-dependently, ~11-15% choline->TMAO (O-103), significant only at >=2 yolks and plateauing 4-6
# (O-104), as an ACUTE POSTPRANDIAL excursion (peak 6-8 h), not a fasting steady-state change (O-105); all 6
# subjects convert with ~4-fold spread (O-106); and no hsCRP/oxLDL change accompanies the rise (O-107).
# Members in HC-9.hypotheses order [H-28 form-governs, H-29 fasting-null, H-30 renal-only,
# H-42 real-postprandial-rise, H-43 benign, H-56 residual]. Anchored on H-43 = 1.0 (rule 7).
lik_miller_H43 = 1.0 # anchor / best fit: H-43 ("the rise occurs but is benign — postprandial, no
# inflammatory-oxidative change") predicts the WHOLE pattern, uniquely including O-105
# (postprandial, not fasting) and O-107 (hsCRP/oxLDL flat). A-40 (approved): the
# postprandial spike with unchanged markers does not establish harm.
lik_miller_H42 = 0.9 # 0.9x: "eggs (>=2 yolks) raise TMAO production" matches O-103/O-104 (the >=2-yolk
# threshold is H-42's exact wording; A-39 approved on saturation) and O-106; priced just
# below H-43 only because H-42 is silent on the postprandial/benign half the full
# pattern displays (consistent, not predicted).
lik_miller_H29 = 0.65 # 0.65x: a postprandial-only flux against a fasting null is COMPATIBLE with H-29 (per
# E-48, O-105 is the reconciling discriminator — it does not contradict the fasting
# null); but H-29 does not itself predict a dose-dependent rise, so well below the poles.
lik_miller_H30 = 0.5 # 0.5x: a postprandial PRODUCTION rise in (normal-renal) healthy subjects is neither
# predicted nor contradicted by a renal-clearance-modifies-steady-state story, which
# governs accumulation rather than single-meal flux — middling.
lik_miller_H28 = 0.3 # 0.3x: egg choline demonstrably converting to TMAO (11-15%, dose-dependent, every
# subject) is against H-28's "egg-PC poorly converted / negligible" reading (E-46/E-53);
# not lower because 11-15% is modest vs free choline, leaving a weak "poor" defence.
lik_miller_H56 = 0.5 # 0.5x residual, middling (rule 3): the ~4-fold host/sex conversion spread (O-106) is
# the person-specific effect H-56 would emphasise, but O-107's flat harm markers argue
# mildly against a clinically-important other relationship. Unconstrained -> mid.
t_miller = 0.73 # cap = min trust_score over {O-103..O-107}, all via S-47 = 0.73; kept at cap — the tiny
# n=6 is already priced into S-47's trust_score (caps precision, not the direction of
# the acute rise, which the wider TMAO literature corroborates).
evidence("HC-9", ["O-103", "O-104", "O-105", "O-106", "O-107"], [lik_miller_H28, lik_miller_H29, lik_miller_H30, lik_miller_H42, lik_miller_H43, lik_miller_H56], t=t_miller)