top-100 methodology + provenance review v1

thesis -> mechanism -> lever -> one step

1) executive read

что у нас уже было

что стало ясно при пересмотре

что получилось

2) where the current logic actually came from

A. guideline / official evidence layer

главные anchors лежат в:
- ai/data/sources.md
- ai/data/panel_evidence_matrix_v1.csv

что там уже strong:
- lipids / apoB / Lp(a) / CAC:
- C-101, C-105, C-122, C-131
- dysglycemia / glucose / HbA1c / OGTT:
- C-102
- kidney / eGFR / albuminuria:
- C-103, C-104
- thyroid caution / repeat logic:
- C-118
- vitamin D caution:
- C-119
- iron / ferritin:
- C-120
- liver fibrosis gate (FIB-4):
- C-121
- imaging boundaries:
- C-107, C-108, C-123, C-124, C-125, C-126, C-127, C-128
- vascular mechanics / PWV:
- C-133, C-134
- VO2max:
- C-129

B. expert / operational layer

главные anchors:
- ai/context/stanislav_skakun_biodata_call_2026_02_28_v1.md
- ai/data/first_panel_decision_packet_v1.md

что оттуда пришло:
- phased MVP instead of full-checkup from day 1
- baseline anchor:
- cbc
- expanded biochemistry
- hs-crp
- insulin / hba1c / glucose context
- tsh
- sex hormone block
- ultrasound-first and partner MRI/endoscopy
- reserve sample / reflex logic
- cIMT as adjunct, not standalone trigger

C. founder / cohort-fit layer

главный anchor:
- ai/context/founder_input_2026_02_22_25_v1.md

что оттуда пришло:
- target cohort = high-agency founders / operators
- raw data + API + provenance > black-box score
- core biomarkers ~100-130
- skin baseline + ecg/vitals + optional dexa/cgm
- "money is not primary"
- machine-readable longitudinal self-use as filter

D. health-system fit layer

главный anchor:
- ai/data/top100_blood_urine_health_audit_v1.md

что оттуда пришло:
- generic top-100 надо сверять с реальной Health truth-layer
- urine block оказался главным фактическим debt
- igf1 surfaced as serious add-candidate
- generic design было полезно, но не до конца закрывало endocrine/adrenal reading

E. equipment / companion-module doctrine

anchors:
- ai/data/mvp_equipment_philosophy_v1.md
- ai/data/vascular_intelligence_layer_v1.md
- ai/data/vascular_intelligence_ops_v1.md

что оттуда пришло:
- PWV как core vascular mechanics signal
- skin / dermoscopy / face / hair as machine-readable longitudinal companion layer
- hair as real endocrine / stress / nutrient-response surface, not cosmetic garnish

3) methodology evolution: v1 -> v2

layer v1 v2
cohort generic preventive mostly healthy high-stress / high-travel founders, men + women
score bps v1 included cost_efficiency bps v2 removed price
count biomarkers + derivations + non-lab items were too mixed exact count = only orderable lab biomarkers
system-age present but lighter endocrine weight endocrine / recovery materially upgraded
hpa read too flat / under-specified dynamic read closed explicitly
vascular blood + imaging boundary present blood + PWV + carotid + CAC tied into one loop
skin/hair present in equipment doctrine now treated as required companion layer

4) provenance map by cluster

cluster why it is in the model primary provenance status
cbc / hematology baseline oxygen / marrow / inflammation context framework logic + evidence matrix, but no clean hematology claim ids yet medium / gap
cmp / chemistry / liver / electrolytes intervention safety + organ context evidence matrix, but chemistry anchors still partly gap medium / gap
glycemic core strongest preventive longitudinal loop C-102 + framework + system-age model strong
kidney / urine / uacr / cystatin silent drift + serum-only miss protection C-103, C-104 + audit strong
lipids / apoB / Lp(a) / CAC causal vascular burden + reclassification C-101, C-105, C-122, C-131 strong
thyroid baseline sanity + repeat caution C-118 strong for caution, medium for broad default use
iron / ferritin / TSAT common false-action zone; needs context C-120 + framework logic medium-strong
vitamin D common panel item but not strong default-screening evidence C-119 cautionary strong
liver fibrosis gate cheap escalation logic C-121 strong
imaging exclusions avoid prestige/incidental theater C-107, C-108, C-123, C-124, C-125, C-126, C-127, C-128 strong
VO2max powerful fitness/mortality signal C-129 + expert/founder tension strong signal, operationally mixed
vascular mechanics / PWV closes vascular age loop beyond lipids alone C-133, C-134 + vascular docs strong
skin / melanoma baseline companion longitudinal surface layer founder + equipment doctrine; repo has cautionary C-106, but not a strong routine-screening support anchor product-strong / evidence-gap
hair / trichoscopy baseline companion endocrine/stress-visible layer equipment doctrine + founder fit; no formal source registry anchor yet product-strong / evidence-gap
endocrine / androgen closure required for useful free-T and recovery read founder fit + Health fit + framework v2; repo lacks strong direct claim anchors important / evidence-gap
adrenal / HPA / SNS closure critical for this cohort, absent from v1 logic Health fit + framework v2; repo currently lacks strong source-registry anchors for spot urinary cortisol, urinary metanephrines, 3-methoxytyramine important / evidence-gap
GlycA / holoTC / IGF-1 / apoE / NMR-LDL-P / sdLDL promoted in v2 because they sharpen system-age or discordance reads mainly v2 product/system logic; not yet fully backed by repo claim ids medium / gap

5) what we were really relying on

strong-evidence zones

medium zones

weak or missing-by-repo zones

6) what v2 changed because of cohort fit

explicit additions

explicit closure rules

explicit counting cleanup

7) the HPA block we locked now

best-read block:
- cortisol_am
- cortisol_pm
- spot_urinary_cortisol_creatinine
- urinary_metanephrines_fractionated
- urinary_3_methoxytyramine

why this is the chosen shape:
- am + pm gives diurnal shape
- spot urinary cortisol gives a more integrated read without 24h collection fragility
- urinary metanephrines separate sympathetic / catecholamine load from pure cortisol story
- 3-methoxytyramine adds extra catecholamine-metabolite nuance when we want the fullest read

what we explicitly rejected as default:
- 24h urinary free cortisol

why:
- too much collection friction
- weak routine longitudinal compliance for this cohort
- bad default if the real product is repeatable tracking

8) what came out of the review

the good

the weak

the actual result now

9) what must be added next

highest-priority provenance debt

  1. HPA / SNS evidence anchors:
    - spot urinary cortisol / creatinine
    - urinary metanephrines fractionated
    - urinary 3-methoxytyramine
  2. endocrine / recovery anchors:
    - IGF-1
    - free T interpretation logic in preventive cohort
    - DHT / AMH / progesterone timing rationale
  3. v2 promoted markers:
    - GlycA
    - holoTC
    - NMR-LDL-P
    - small dense LDL
    - apoE baseline-consent framing
  4. companion-module provenance:
    - skin mapping / dermoscopy as longitudinal protocol object
    - hair diagnostics as endocrine/stress-response longitudinal object

operational debt

10) final read

source map inside repo