BoardsBridge

BoardsBridge · Editorial Process

How we review every ABA rapid card

Clinical accuracy on a fast-moving specialty exam requires structure, not vibes. Every rapid card on this platform passes through the same editorial process and the same review criteria, before it ships and on a yearly cadence thereafter.

v2023 ASA Ethics
ASRA LAST 2020
AHA ALS 2025
MHAUS 2018/2019

Editorial process

Cards are reviewed by a licensed physician (the accountable signer) with cross-source verification by frontier-class language models. We call this an editorial process rather than a "committee" — it is a single accountable MD using verification tooling, not a peer deliberation body. Each card is signed individually; nothing on this platform is the product of any majority-of-MDs process.

  • Accountable signer: Uri Rozen, MD — Licensed Physician, Israel
  • Verification layer:
    • Claude Opus 5
    • GPT 6 Astra
    • Gemini 3.7
    • Proprietary in-house trained models
  • Process: Draft → cross-source verification (model-assisted) → human sign-off → build-time Zod validation (schema + section completeness + ≥ 2 manifest-resolvable guideline citations) → ship to src/content/aba-rapid-cards/. Each card's review timestamp is stored in seo.reviewedAt.

No model output is published without MD sign-off. No claim is published without a resolvable source. The loader enforces both as build errors — the build will fail before a card with insufficient review or with fabricated citations reaches production.

Editorial criteria

  1. Doses are within current guideline range. Manual cross-check + automated verification against the cited source guideline.
  2. Mechanism is aligned with cited sources. No fabrication, no extrapolation beyond cited evidence.
  3. No fabricated citations. Every reference resolves in content/ABA/guidelines/sources.manifest.json or carries the literature: prefix for textbook citations.
  4. Board trap is real, not trivia. Reflects the classic examiner follow-up the resident might miss — not a filler.
  5. If-you-forget is the one line. Single, high-yield, anchor. Not a paragraph.
  6. Recency. All cited guidelines are ≤ 5 years old unless explicitly flagged as historical.
  7. Guideline density. Each card cites ≥ 2 authoritative guideline sources (manifest-resolvable). This is the strongest E-E-A-T signal available to us.

Cadence & triggers

Re-review cadence: Cards are re-reviewed on a per-topic-type cadence. Crisis-management content (the high-stakes, frequently-updated section) is re-reviewed every 6 months; everything else every 12 months. Cards approaching or exceeding their cadence window are blocked from the public SSG build until re-reviewed.

Earlier re-review triggered by:

  • source guideline update
  • high severity feedback

Source guidelines we anchor to (see sources.manifest.json): MHAUS, ASRA, AHA, SOAP, ASA, SCCM, ACC/AHA, KDIGO, ERNICA, EACTS, NICE, ENETS, equivalent bodies.

Disclaimer

Educational reference only. Reviewed for clinical accuracy against current authoritative guidelines. Not a substitute for clinical judgment, institutional protocols, or current guidelines. Every card carries this disclaimer in a visible banner above the clinical content.

We model the structure of AI-assisted medical content products (Glass Health, OpenEvidence, Hippocratic AI). The physician on the editorial process is the accountable signer; the verification layer is a tool, not a decision-maker.

BoardsBridge Editorial Process · 12-month review cadence · last-updated 2026-09-10

Feedback

Found an error on a card? Spot a stale dose or a missing reference? Email clinical-review@boardbridge.live with the card URL and the issue. High-severity reports trigger an immediate re-review under our high severity feedback policy.