Every approach to the same diagnosis, side by side — all categories load from the Layer-1 knowledge file. Two-axis tiers per the CDS Evidence Schema v2 + the 08-22 vetting memo. Every row cites its sources.
Draft reference: clinical review, scope and source notes
⚠ MODEL-DRAFTED SCAFFOLD — every tier is provisional pending Brian's clinical review (Schema v2 rule). Not patient-facing, not a protocol, not marketing. Doses are literature/study doses, not prescriptions. REG tiers: GREEN = lawful today · Y1–Y3 = pathway-gated (see External Vetting Memo §3) · RED = off menu. All three traditional-medicine lanes (Ayurvedic · Chinese herbal · Western/ancient) populated 2026-08-22 from citation-verified literature sweeps (~90 sources; PMID/DOI per row). Standing rule for herbal rows: structure/function language only in any patient-facing use; herb-drug interaction check documented at enrollment; hold blood-moving herbs ≥7 days pre-procedure. Transposition disclosure (mechanism pass, 08-22): where an agent appears under a diagnosis without direct trials there, its tier badge reflects its best-evidenced indication and the use is a labeled mechanistic transposition per Evidence Schema v2 — the Mechanisms tab shows the mapping and rationale per diagnosis.
Severity = clinical priority for THIS practice's population: HIGH act (avoid / hold / substitute / monitor tightly) · MED counsel + monitor · LOW awareness/spacing. Class-level rules — the row's action governs any member of the med class. PROVISIONAL until Brian's review.
Before a procedure, review connected screening and current procedure-specific guidance: kidney function, antithrombotics, infection risk and glucose where indicated. The legacy timing table below is not automatic clearance and has not been fully revalidated in this update.
Defaults from ASRA interventional-pain anticoagulation guidelines, 2nd ed (Narouze et al., Reg Anesth Pain Med 2018) + practice rules from the 08-22 herbal sweeps. ⚠ NEVER hold antiplatelets after recent coronary stent (DES <6–12 mo) without cardiology sign-off. Hold/restart decisions coordinate with the prescriber; risk tier is per-procedure AND per-patient. PROVISIONAL — Brian confirms tiers + practice policy rows before deployment.
The practice formulary/arsenal, grouped, with each group's top interaction axes. PROVISIONAL roster drafted from practice records — Brian confirms the formulary.
CLINICAL MANUAL — the teaching layer. Written for a medical audience: selection rationale, pharmacology, study-level evidence with certainty grades, adverse-effect profile, interactions, dosing as studied, and an explicit EVIDENCE GAPS section on every entry — where rigor is missing, the entry says so rather than papering over it. Select a diagnosis, then a discipline, then click any agent to open its monograph. Entries marked CONDENSED carry the atlas summary only; full monographs pending. ⚠ MODEL-DRAFTED from the cited sources — physician editorial review converts this to teaching canon; until then it is a draft manual, not doctrine.
MECHANISMS & EDUCATION — the foundational teaching layer: WHY each diagnosis hurts (the disease mechanisms) and WHY each recommended therapy maps to those mechanisms. Two audiences per diagnosis: STAFF (pathophysiology + mechanism→therapy map, for training Maywell employees on why we recommend what we recommend) and PATIENT (plain-language explanation, suitable as handout source text after physician review). ⚠ MODEL-DRAFTED — physician editorial review required before any patient-facing use; patient text uses honest, non-promissory language by design.
Staff versionPatient version
v120 layers. DOCTRINE is the atlas's own principles expressed as data; WARRANT is the axis that decides what earns a place when the evidence isn't there. The gate report below runs the principles against all rows live in this page — it is the same logic as doctrine_check.js, so what you see here is what blocks a mint.