LAYER-2 MATCHING ENGINE — v0.4 TEST HARNESS · DRAFT, NOT CLINICAL GUIDANCE · reads LIVE from atlas-data.js (Layer 1: 17 categories incl. the 2026-08-28 vertical-scan builds) · payer-lane aware · tiers provisional until confirmed
Test environment · engine under evaluation · not clinical use · stateless — nothing is saved
Layer 2 — Matching Engine Test Harness v0.2
Live against the real atlas (— rows, census 2026-08-27).
Built to make the logic fail visibly.
Intake
0answered
0from note
0remaining
0in scope
Clinical noteproposes only
DiagnosisF1
Stage 0 — eliminators12
Fail-safe mode — when a row carries no structured safety data and a bleeding-relevant
eliminator is active, treat it as GATED, not open. 14 of 17 procedural rows currently
have no safety field, so absent prose otherwise reads as absent risk.
Fail safe (recommended)
Payer lane — WC mode GATES rows whose Layer-1 notes carry a WC Not-Recommended /
never-bill flag (prose-derived until the structured wc_lane field lands; a row
with no flag is NOT cleared for WC — absence of a flag is a data gap).
Polysymptomatic distress (0–31) — the centralization modifier. Each point raised the
odds of failing an analgesic-response threshold by ~18% after joint replacement, continuously and
well below the fibromyalgia cutoff. Re-weights the ranking; never gates care.
0 · Low
Standard pathway.
TRANSPOSED: predictive evidence is arthroplasty and spine surgery — not ESI/MBB/RFA.
Stage 3 — willingness—
Only families present in this diagnosis are asked about. 0 = refuses → hard filter F5.
Stage 4 — burden5
Pills/day tolerated
Monthly out-of-pocket ceiling
Visits per month tolerated
Preparation effort tolerated
Routine screeners22 · already collected
These you already run. Marginal cost zero. PEG and ODI
carry phenotype; GAD-7 and PHQ feed the support profile only.
Item text not reproduced. PSEQ-2, CEQ-expectancy, Brief IPQ and BMQ
are licensed or license-ambiguous. The harness models their scores so the flow and the engine
can be tested; paste real items in after the licensing check.
Nec−concerns50
Expectancy60
Self-efficacy40
Only necessity−concerns reaches the ranking function. The other two are
structurally absent from scoring — see the engine source.