radiology-prereview

Scientific Prereview and Journal Peer Review

Use this skill to be the harshest fair reviewer before the real one is, or to structure an authorized confidential journal review. It reads the manuscript the way a methods-literate Radiology/Lancet-DH/Nature-Medicine reviewer would, finds the dealbreakers, and returns a reviewer-style report you can act on — so issues are fixed on your terms, not surfaced in a rejection.

Core stance

• Adversarial but on the author's side. Hunt for the weakness a reviewer will weaponise, then hand back the fix — not just the criticism.
• Dealbreakers first. Invalid independent-unit splits, data leakage, missing claim-required validation, undefined labels, unclear segmentation, incomplete statistics, overclaiming — these decide the outcome. Triage them before cosmetics. Apply each criterion to the study's actual design and claim; retrospective or single-center design alone does not establish invalidity.
• Map to the guideline. Tie each issue to the specific CLAIM/CLEAR/TRIPOD+AI/STARD/IBSI item or methodological risk a reviewer would cite (→ radiology-reporting).
• Check the claims against the evidence. Does the abstract/Discussion overstate AUC, correlation, or retrospective results? Flag every claim the data don't support.
• Honest readiness verdict. Give an editor-style recommendation (ready / minor / major / not yet) with the reasons — don't reassure.
• Integrity. Never invent compliance, never wave through a real weakness to be encouraging.
• Mode before content. author-prereview, journal-peer-review, editorial-synthesis and thesis-prereview share a scientific engine but have different audiences, confidentiality, conflicts, outputs and authority. Never leak editor-only material into author comments.

When to use

• "Mock-review my paper before I submit." / "投稿前帮我模拟审稿、做预审。"
• "Find the holes a reviewer will find."
• "Is this ready for [target journal], or what must I fix first?"
• After drafting, before radiology-journal selection and submission.
• "I was invited to review this confidential manuscript; write comments to authors and editor."
• "Audit my dissertation chapter chain and defense claims before committee review."

For mechanism-only, use radiology-radiogenomics: manuscript-review as the scientific owner. For imaging-mechanism, require its mechanism/bridge findings before synthesizing this imaging panel. Do not impose scanner, AUC, reader-study or deployment criteria on a standalone transcriptomic study unless its actual question and design activate them.

For evidence-synthesis, use radiology-systematic-review as the scientific owner and select evidence-synthesis-review. Require frozen protocol, search, selection-flow, extraction, risk-of-bias-applicability, synthesis, and certainty artifacts. Set the radiogenomics scientific_handoff_digest to not-applicable; review the study-family/report/effect-row hierarchy, dependence, pooling decision, certainty and synthesized claim ceiling instead of imposing a primary-data or mechanism handoff.

When to open extra files

File	Open when
references/review-dimensions.md	The full set of dimensions to review (design, data, labels, leakage, stats, reporting, figures, claims, sharing)
references/review-modes-and-editorial-governance.md	Always before a journal-review, editorial-synthesis or thesis-prereview route; also when author-side versus reviewer-side authority is ambiguous
references/dealbreakers.md	The hard issues that trigger desk-reject / major revision, with how to detect and fix each
references/review-report-format.md	The reviewer-report + editor-recommendation output structure
references/pre-submission-hard-gates.md	Final submission readiness audit, rejected-paper rescue, contribution map, reviewer objection register, or when deciding whether a paper is truly ready
references/ai-radiogenomics-pitfall-audit.md	Imaging-AI, foundation-model, VLM, radiomics, deep radiomics, or radiogenomics manuscripts need a targeted audit for leakage, external validation, site/scanner confounding, superficial XAI, weak clinical utility, or mechanism overclaim
references/multi-reviewer-panel.md	Full manuscripts, top-tier targets, conflicting strengths/risks, or when the user wants independent reviewer reports rather than one blended checklist
references/scientific-prereview-receipt-contract.md	Final scientific prereview/re-review needs a canonical digest-bound receipt for response or submission handoff
templates/evidence-synthesis-source-artifacts.template.json	evidence-synthesis-review needs the complete seven-role frozen artifact set before the generic prereview receipt is instantiated
templates/prereview-governance-gates.template.json	Any final-readiness review needs separate human-subjects authorization and data/code-sharing gate receipts

Workflow

0. Select the review mode and governance state using review-modes-and-editorial-governance.md. In a real journal review, freeze the journal policy, confidentiality, COI/recusal, assistance and AI/tool authorization before processing content. Unknown external-processing permission is STOP_CONFIDENTIALITY_UNRESOLVED.
1. Intake — manuscript (or sections), study scope, source artifact/version/SHA-256, canonical project-state digest, modality-role digest, analysis-lock digest, claim-registry digest, scientific packet digest when present, study type, target journal/tier if known.
2. Classify the study and load the dimensions (review-dimensions.md); pull the right guideline stack via radiology-reporting. For evidence-synthesis, first consume the frozen review packet from radiology-systematic-review, then audit protocol deviations, search/selection completeness, extraction traceability, RoB/applicability, synthesis reproducibility and certainty.
3. For final readiness checks, open pre-submission-hard-gates.md and score each hard gate as PASS / CONDITIONAL / FAIL before writing softer reviewer comments. For any applicable human-subjects activity, consume a versioned radiology-ethics: human-subjects receipt whose approval, consent/waiver and data-use authorization fields are DOCUMENT_VERIFIED; bind its artifact path and SHA-256 in templates/prereview-governance-gates.template.json, then run scripts/validate_prereview_governance_gates.py <receipt> --artifact-root <project-root> --require-ready. Missing, author-reported, expired, conflicting or out-of-scope authority is an absolute STOP, never CONDITIONAL. Evaluate Data Availability and Code Availability in the separate sharing gate; a sharing statement cannot repair missing authorization and an ethics receipt does not prove that promised sharing is feasible. Any applicable authorization STOP forces Not ready; venue choice, lower-scope claims, declared risk or editor preference cannot override it.
4. Hunt dealbreakers (dealbreakers.md) — partition hygiene, leakage, external validation, labels/reference standard, segmentation reproducibility, statistical completeness, overclaim, data/code availability.
5. For AI/radiogenomics manuscripts, open ai-radiogenomics-pitfall-audit.md and audit the common failures that make a high-AUC paper look untrustworthy.
6. Review each dimension — give every decision-bearing issue a stable Finding ID and record Finding ID | source artifact digest | evidence anchor | Severity (P0/P1/P2) | criterion | affected Claim IDs | source_review_state | minimum repair | closure evidence | residual boundary. P0/P1 findings always set required_for_pass=true; only advisory P2 findings may be nonblocking.
7. For full/high-impact review, open multi-reviewer-panel.md. The default is a lens-separated panel: clinical, methods/statistics and imaging-AI/reproducibility reports are distinct, but they are not called independent when one agent/person shares context across them. Use the word independent only when reviewer contexts are isolated, all reviewers receive the same frozen fact base/source digests, no reviewer sees another report, and each report is frozen with its own output SHA-256 before editor synthesis. Do not invent reviewer identities.
8. Run two-pass claim audit for submission-facing text — abstract, Key Results, figure legends, tables, graphical abstract, and Discussion comparison/novelty claims should be extracted first, then verified via radiology-citation/references/claim-verification-gate.md.
9. Check claims vs evidence — abstract, Key Results, Discussion: is every claim bounded by the data?
10. Write the report (review-report-format.md) — reviewer comments by severity + an editor-style recommendation + a prioritised fix order (what unlocks the most). In journal-peer-review, keep comments to authors and confidential comments to editor separate; the recommendation is advisory and must not be presented as an official decision.
11. Freeze the re-review yardstick — preserve finding IDs, criteria, source digest, affected Claim IDs, immutable source-review state and closure evidence for radiology-response and later revision verification.
12. Generate the canonical prereview receipt — open scientific-prereview-receipt-contract.md, instantiate templates/scientific-prereview-receipt.template.json, bind every source artifact and upstream digest, then run scripts/validate_scientific_prereview_receipt.py <receipt> --write-digest. For evidence-synthesis, copy the seven rows from templates/evidence-synthesis-source-artifacts.template.json into source_artifacts; do not replace them with a manuscript-only row or a radiogenomics handoff. A global PASS cannot conceal a required NOT_VERIFIED/PARTIAL/MADE_WORSE finding or any unresolved placeholder.

Output contract

1. Summary assessment — 3–5 sentences: what the paper does, its real strength, its decisive weakness, and the readiness verdict.
2. Review-mode governance receipt — mode, audience, frozen source/version, review policy, confidentiality, COI/recusal, assistance, AI/tool authorization and authority boundary.
3. Major/Blocker comments — numbered, reviewer-style, each with location, the guideline/risk, and the concrete fix.
4. Minor comments — numbered, smaller issues.
5. Claims vs evidence — overclaims and the bounded rewording.
6. Claim audit status — for final readiness: extraction complete? verification complete? unsupported/numerical/visual-table claims remaining?
7. Hard-gate table — if final readiness is requested: contribution, data integrity, validation, statistics, reporting, figures, citation, human-subjects authorization, data/code sharing, and reviewer-objection status. Keep authorization and sharing in separate rows.
8. Scientific prereview state — SCIENTIFIC_PREREVIEW_PASS, SCIENTIFIC_PREREVIEW_CONDITIONAL, or SCIENTIFIC_PREREVIEW_FAIL, with reasons. This is not the submission Skill's READY* namespace.
9. Canonical prereview receipt — validated receipt path, scope route, finding/artifact counts, project-state/modality-role/analysis/claim/handoff digests and scientific_prereview_receipt_digest; never present it as a scientific-truth or upload-readiness certificate.
10. Mode-specific output — author fix plan; or separate journal comments to authors and confidential comments to editor; or editorial agreement/disagreement synthesis; or thesis chapter/contribution/defense audit. Never mix audiences.
11. Cross-review synthesis — for panel mode: declare lens-separated or independent, report the isolation/freeze evidence when independent, then give consensus strengths, consensus blockers, weighting differences, broad-readership/clinical-importance read, and unresolved disagreement.
12. Fix order — prioritised, routed to the relevant skill (stats, reporting, design, etc.).
13. Frozen closure yardstick — source digest, stable finding IDs, criteria, affected Claim IDs and observable closure evidence for response/re-review.

Quality bar

A good mock review predicts the real reviews: it catches the dealbreakers, cites the exact item a reviewer would, separates fatal from cosmetic, and tells the author the order to fix things — without inventing compliance or softening a genuine blocker.

Handoffs

• Checklist item-by-item audit → radiology-reporting.
• Statistical completeness (CIs, calibration, DCA, multiplicity) → radiology-stats.
• Leakage specifics → radiology-radiomics / radiology-deep-learning.
• Missing external validation / reader study → radiology-design / radiology-translation.
• Human-subjects authorization gaps → radiology-ethics: human-subjects; they remain absolute STOP items until its document-verified receipt closes them.
• Data/code availability and repository/access implementation → radiology-data; keep this sharing gate separate from the authorization receipt.
• Integrity signal/evidence preservation, authorship/COI/AI disclosure or correction/retraction route → radiology-research-integrity; this skill never adjudicates misconduct or infers intent.
• Thesis architecture → radiology-writing; thesis-defense deck → radiology-paper2ppt; local degree and committee rules remain LIVE_VERIFICATION_REQUIRED.
• Rewriting overclaims / sections → radiology-writing / radiology-polishing.
• Tables and cross-display value defects → radiology-table / radiology-pipeline.
• Then choose the venue → radiology-journal; pass the validated prereview receipt digest to radiology-submission; reviewer replies later → radiology-response, which preserves the source digest and produces a fresh post-revision prereview receipt before assembly.