biomarker top-100 selection framework v2

thesis -> mechanism -> lever -> one step

• thesis: новый top-100 нужен не для "красивого длинного чекапа", а для максимально плотного longitudinal signal в Health для high-stakes / high-stress / travel-heavy cohort.
• mechanism: пересчитываем v1 через bps v2, убираем цену из формулы, усиливаем system-age fit, отделяем orderable biomarkers от derivations и от companion phenotyping modules.
• lever: каждый order-marker обязан либо закрывать системный age-loop, либо разрешать реальный discordance/decision gap.
• one step: после research-pass прогнать v2 rerun, выдать shared core + sex overlays + gated lane + companion modules, и только потом lock exact top-100.

1) what changed vs v1

• cohort shift:
• v1: mostly healthy generic preventive cohort.
• v2: entrepreneurs / operators under chronic cognitive load, travel stress, sleep disruption, sympathetic overdrive, and hormone-recovery fragility.
• scoring shift:
• cost_efficiency removed from the priority formula.
• price is explicitly out of the decision loop unless it blocks assay quality or method provenance.
• product shift:
• value is not "bigger panel".
• value is system-age coverage + retest discipline + machine-readable longitudinal comparability.
• architecture shift:
• exact top-100 count applies to orderable lab biomarkers only.
• derivations and non-lab modules are mandatory layers, but they do not consume slots inside the 100-count.

2) high-stress cohort doctrine

• primary cohort:
• men and women with high decision load, high travel load, compressed recovery windows, and willingness to measure repeatedly.
• default interpretation stance:
• trajectory optimization, resilience, and hidden deterioration prevention.
• default blind spots we must close:
• androgen status is unreadable without total_testosterone + SHBG + albumin -> calculated free T.
• adrenal / HPA load is unreadable from cortisol_am alone.
• vascular aging is incomplete without mechanics (PWV) in addition to blood and imaging.
• visible-aging / surface layer is incomplete without skin and hair baselines.

3) hard gates v2

no candidate enters core unless all of the following are true:
1. valid assay or acquisition method with clear provenance.
2. explicit path measure -> threshold/trend -> action -> retest.
3. known pre-analytic SOP and discordance policy.
4. claim language is safe for asymptomatic preventive use.
5. marker fits the high-stress achiever cohort.
6. marker improves at least one system-age loop or resolves a critical ambiguity inside it.

4) biomarker priority score (bps v2)

bps_v2 = 0.35*evidence + 0.25*actionability + 0.20*reproducibility + 0.10*longitudinal_signal + 0.10*operational_fit

scale per axis: 1..5

status mapping:
- >=4.3: core
- 3.5..4.29: gated
- <3.5: research/hold

note:
- operational_fit stays in the model because broken logistics still destroy longitudinal truth.
- price is not a reason to demote a marker if the method is defensible and the longitudinal value is real.

5) exact-count discipline

three different objects must not be mixed:

A. orderable biomarkers

• real assays we order and collect.
• only these count toward the exact top-100.

B. derivations

• examples: free_testosterone_calc, homa_ir, uacr, fib4, non_hdl_c, apob_apoa1, tsat.
• these are required outputs, but they do not count as separate order markers.

C. companion phenotyping modules

• vascular mechanics, skin mapping, dermoscopy, face skin imaging, hair/scalp quantification, body composition, ECG, VO2max.
• these are mandatory parts of the longitudinal operating system, but they are not "biomarkers" inside the 100-count.

6) mandatory v2 promotions from v1

these candidates must be in the next rerun universe, not left as side notes:
- free_t3
- fasting_c_peptide
- igf_1
- cortisol_pm
- spot_urinary_cortisol_creatinine
- urinary_metanephrines_fractionated
- urinary_3_methoxytyramine
- glyca
- holotranscobalamin (holoTC) or explicit dual-track with total B12
- apoe_genotype as one-time baseline candidate
- apoa1
- nmr_ldl_particle_number
- small_dense_ldl

7) endocrine / adrenal closure is mandatory in v2

testosterone closure

minimum male/female androgen-read requirements:
- total_testosterone
- shbg
- albumin
- derived free_testosterone_calc
- estradiol
- lh
- fsh

male-specific add:
- dht

female-specific core overlay:
- estradiol day 2-5
- progesterone luteal
- amh

adrenal / hpa / sns closure

minimum v2 baseline must read both snapshot and integrated load:
- cortisol_am
- cortisol_pm
- dhea_s
- spot_urinary_cortisol_creatinine
- urinary_metanephrines_fractionated
- urinary_3_methoxytyramine
- prolactin
- igf_1

why:
- one morning cortisol point is too flat for this cohort.
- high-stress / high-travel users need HPA shape, not only adrenal-insufficiency screening.
- 24h_urinary_free_cortisol is not default baseline because collection friction is too high; keep it only as a fallback / confirmatory path when spot urine read is unclear or clinically necessary.

8) canonical discordance reads for v2

no single-marker protocol change unless guardrail-level danger exists.

default v2 discordance rules:
1. glucose vs hba1c
2. creatinine vs cystatin_c
3. ferritin vs hs_crp/glyca
4. tsh vs free_t4 vs free_t3
5. total_testosterone vs shbg vs free_testosterone_calc
6. ldl_c vs apob vs nmr_ldl_particle_number
7. cortisol_am vs cortisol_pm vs urinary_cortisol vs urinary_metanephrines

9) companion modules that must travel with the rerun

vascular intelligence companion layer

• blood core:
• apoB, Lp(a), lipid context, inflammation, blood pressure
• mechanics core:
• PWV is the default vascular-elasticity signal
• structural adjunct:
• plaque-first carotid ultrasound
• gated adjunct:
• CAC
• reserve / gated:
• FMD

rule:
- vascular elasticity is represented by PWV, not by vague "vascular age" language.

skin / melanoma / surface layer

• total body skin mapping
• dermoscopy workflow for suspicious lesions
• longitudinal lesion-change comparison
• face skin imaging

rule:
- skin surface data is part of the preventive longitudinal operating system, not cosmetic fluff.

hair / scalp layer

• digital trichoscopy
• density, shaft thickness, anagen/telogen ratio, zone map

rule:
- hair is a visible endocrine / stress / nutrient-response layer and belongs in the companion baseline.

10) system-age fit rule

a marker gets promoted when it improves one of these loops:
- cardio-vascular age
- glyco-metabolic age
- renal age
- liver-metabolic age
- inflammation / immune age
- endocrine / recovery age
- body-composition / fitness age

special v2 stance:
- endocrine / recovery age gets materially more weight than in v1.
- visible-aging layers (skin, hair) are tracked as companion signals and should not be faked into the blood-only score until calibration is explicit.

11) what stays out even without a price gate

• broad shotgun tumor-marker bundles in asymptomatic users
• routine whole-body CT / MRI as default screening
• microbiome panels without method stability / comparability
• consumer epigenetic-age products without transparent method and reproducibility
• standalone cIMT as universal default trigger

12) outputs required from the rerun

1. exact v2 top-100 orderable biomarker list.
2. explicit shared core.
3. explicit male overlay core.
4. explicit female overlay core.
5. explicit gated lane.
6. derivation list kept outside the 100-count.
7. companion module list (vascular, skin, hair, body-composition, ECG, VO2max).
8. per-marker role tags: guardrail | response | engagement.
9. cadence + retest + discordance rules.

13) known open questions

• holoTC as full replacement vs reflex add-on to total B12.
• glyca as supplement vs partial replacement for hs_crp.
• apoE consent and downstream communication protocol.
• whether OGTT + insulin curve should remain gated or move to wider default use in this cohort.
• whether visible-aging layers should stay separate from overall_age_delta until enough repeat data exists.