medsci-skills 5.10.0 → 5.11.0

package/metadata/distribution_files.json

@@ -1658,8 +1658,8 @@
     },
     {
       "path": "skills/humanize/SKILL.md",
-      "size": 13461,
-      "sha256": "78e12b996cacfa5b1b16b0a0fe26cee5dd72cb990b70f26e90041d6e0eb65c9a"
+      "size": 14240,
+      "sha256": "f179598c74b1ab0b3d6639fdd26b8cafa33713a9ebe9374aad5d31571340d836"
     },
     {
       "path": "skills/humanize/references/ai_patterns.md",
@@ -2923,8 +2923,8 @@
     },
     {
       "path": "skills/peer-review/SKILL.md",
-      "size": 68213,
-      "sha256": "51feccb82dca34a28a0b9ace5467d23ddc654056af2cd9e152a4559cf074cfc3"
+      "size": 68302,
+      "sha256": "bf56fbc16bb3e43fb43f09982d36e68f53f8b0f666564e2828e692472ad0660f"
     },
     {
       "path": "skills/peer-review/references/aczel_2021_reviewer2_patterns.md",
@@ -2933,8 +2933,8 @@
     },
     {
       "path": "skills/peer-review/references/domain-probes/ai_overclaiming.md",
-      "size": 14205,
-      "sha256": "565aa362e20ea6ca923510ea393829fb851e0caaae3d732f99144dd48a25b951"
+      "size": 15410,
+      "sha256": "bdb07002bf0b351579036937af5bf149e35389a218198699b81d31dc2fda7ec9"
     },
     {
       "path": "skills/peer-review/references/domain-probes/case_report.md",
@@ -2948,8 +2948,8 @@
     },
     {
       "path": "skills/peer-review/references/domain-probes/diagnostic_accuracy.md",
-      "size": 9002,
-      "sha256": "8232e2023d3c4d52e6a2d9003ae335302d6fe78c54d39b81f591c07becaf4df0"
+      "size": 11692,
+      "sha256": "afc69e56578babcc1cf7c7952d2df301fdc829329fb480a5cf426ff6ab65d9f5"
     },
     {
       "path": "skills/peer-review/references/domain-probes/equity_fairness.md",
@@ -2993,8 +2993,8 @@
     },
     {
       "path": "skills/peer-review/references/domain-probes/observational_confounding.md",
-      "size": 33429,
-      "sha256": "ac1736047c069aa47df76deea92379cae407043b6989d8d3fc77db8fd84b0031"
+      "size": 34617,
+      "sha256": "ec4a11caed6d604e510437827f2b979fcf43ab1ddea6b298eb95673472b2b459"
     },
     {
       "path": "skills/peer-review/references/domain-probes/polygenic_risk_score.md",
@@ -3028,8 +3028,8 @@
     },
     {
       "path": "skills/peer-review/references/domain-probes/sr_ma.md",
-      "size": 17239,
-      "sha256": "486d569f559f16d62b882f7256ccfe7443e3e2b4cc72a407a5cd046162a98b25"
+      "size": 20154,
+      "sha256": "82ccb4abc1ab1fffaf528268baad6c79e67c9b2ff929c14d7dcccb9677e10ddc"
     },
     {
       "path": "skills/peer-review/references/domain-probes/survey_research.md",
@@ -3488,13 +3488,13 @@
     },
     {
       "path": "skills/self-review/SKILL.md",
-      "size": 102413,
-      "sha256": "97219c57684d4eaced9e35b0968ca09da7731095c186a4e97757b4039f43f2f6"
+      "size": 104028,
+      "sha256": "22bc51c9adcb1fc9ff9120c2650faaa2f5f3a0482fe1fd781e231e2191e21081"
     },
     {
       "path": "skills/self-review/references/domain-probes/ai_overclaiming.md",
-      "size": 14205,
-      "sha256": "565aa362e20ea6ca923510ea393829fb851e0caaae3d732f99144dd48a25b951"
+      "size": 15410,
+      "sha256": "bdb07002bf0b351579036937af5bf149e35389a218198699b81d31dc2fda7ec9"
     },
     {
       "path": "skills/self-review/references/domain-probes/case_report.md",
@@ -3508,8 +3508,8 @@
     },
     {
       "path": "skills/self-review/references/domain-probes/diagnostic_accuracy.md",
-      "size": 9002,
-      "sha256": "8232e2023d3c4d52e6a2d9003ae335302d6fe78c54d39b81f591c07becaf4df0"
+      "size": 11692,
+      "sha256": "afc69e56578babcc1cf7c7952d2df301fdc829329fb480a5cf426ff6ab65d9f5"
     },
     {
       "path": "skills/self-review/references/domain-probes/equity_fairness.md",
@@ -3553,8 +3553,8 @@
     },
     {
       "path": "skills/self-review/references/domain-probes/observational_confounding.md",
-      "size": 33429,
-      "sha256": "ac1736047c069aa47df76deea92379cae407043b6989d8d3fc77db8fd84b0031"
+      "size": 34617,
+      "sha256": "ec4a11caed6d604e510437827f2b979fcf43ab1ddea6b298eb95673472b2b459"
     },
     {
       "path": "skills/self-review/references/domain-probes/polygenic_risk_score.md",
@@ -3588,8 +3588,8 @@
     },
     {
       "path": "skills/self-review/references/domain-probes/sr_ma.md",
-      "size": 17239,
-      "sha256": "486d569f559f16d62b882f7256ccfe7443e3e2b4cc72a407a5cd046162a98b25"
+      "size": 20154,
+      "sha256": "82ccb4abc1ab1fffaf528268baad6c79e67c9b2ff929c14d7dcccb9677e10ddc"
     },
     {
       "path": "skills/self-review/references/domain-probes/survey_research.md",
@@ -3658,8 +3658,8 @@
     },
     {
       "path": "skills/self-review/scripts/check_binning_consistency.py",
-      "size": 19541,
-      "sha256": "e3bf7dd2e0871ce6905abc1d33a26c7afac76a93d184bfe2d431af97d0622f74"
+      "size": 20076,
+      "sha256": "c6467eb4a1d954d67da87e283d8a628604a9ec7558c966f8b20c1b661e23b9c9"
     },
     {
       "path": "skills/self-review/scripts/check_citation_order.py",
@@ -3668,13 +3668,13 @@
     },
     {
       "path": "skills/self-review/scripts/check_claim_artifact.py",
-      "size": 13065,
-      "sha256": "ac29a7e8145152e6641fce8eaf98766cfc943a034db72252cc38538d7ab59db2"
+      "size": 15879,
+      "sha256": "1daa64eaabe493f05ae7e2129a2056b673ec588487d5b6c73f41f10104822935"
     },
     {
       "path": "skills/self-review/scripts/check_classical_style.py",
-      "size": 12210,
-      "sha256": "c973ee8b776f28515439fb185e1254e08e62c2e1410e260f18a824241a331af0"
+      "size": 13124,
+      "sha256": "941d90559cdcdd24f3eba58daf00eb34a7d227c3fe44267b93b65b80b6092989"
     },
     {
       "path": "skills/self-review/scripts/check_cohort_arithmetic.py",
@@ -3691,10 +3691,20 @@
       "size": 22032,
       "sha256": "42e0e9315e1c97ca0f9943213c65ea91d7064773d25287b2577f0386dfb7fc41"
     },
+    {
+      "path": "skills/self-review/scripts/check_emphasis_density.py",
+      "size": 6626,
+      "sha256": "79e517942c1eca4f879fa2b05e6497f93fe6f4d53ed58ca98e240bf5b44c4821"
+    },
+    {
+      "path": "skills/self-review/scripts/check_figure_citation.py",
+      "size": 5298,
+      "sha256": "342bc1126a420111da9c0fe38c6cd68cc45542b0e485202689d7219de9d80712"
+    },
     {
       "path": "skills/self-review/scripts/check_null_calibration.py",
-      "size": 7594,
-      "sha256": "9ddff01722c34efb6ffd757ae762c6ee12f5993bf13b11313c2e20b60b26cab3"
+      "size": 8585,
+      "sha256": "e5bd71c515554b3acbaefd3807e584364a536dc6e1101bec5b524bc26b341309"
     },
     {
       "path": "skills/self-review/scripts/check_panel_diversity.py",
@@ -3716,20 +3726,25 @@
       "size": 15481,
       "sha256": "c9cd8e47cfef1ffcea12ecda5314a71ccd5415f35f4f296addadf53ccb6ceb3b"
     },
+    {
+      "path": "skills/self-review/scripts/check_rounded_delta.py",
+      "size": 7079,
+      "sha256": "2961dd0b25dcd06f202442004e08f2d98c6f2d4c337be3f8b0e3286a59952072"
+    },
     {
       "path": "skills/self-review/scripts/check_scope_coherence.py",
-      "size": 10818,
-      "sha256": "820dfc264c2a4f62c79c0c7123a3e1a8b59a100b89654617a08ff55deeb25a75"
+      "size": 11920,
+      "sha256": "e600e54fcbe308060a8dadd871ec274b6ae19f89456ae0194490e152c0d09a91"
     },
     {
       "path": "skills/self-review/scripts/check_supplement_hygiene.py",
-      "size": 11088,
-      "sha256": "f89027472cdf0258357c3b0f0b0f3fec09b5ea65cc1373292797b818d1acf444"
+      "size": 13579,
+      "sha256": "c61bd5e485d2bce4a6efd7bf2a7980aa0d31ebecb876a7c6895fd12cb6bf7f1c"
     },
     {
       "path": "skills/self-review/skill.yml",
-      "size": 2223,
-      "sha256": "4009f3148776fab2096da3dad8a15e503c5073b4cc66b42c57498948e2040270"
+      "size": 2345,
+      "sha256": "91883d75f33ee873a6ebfbd4d5f2cf0c292d3b90fe4705d6a35e82a26ecf6523"
     },
     {
       "path": "skills/setup-medsci/SKILL.md",

package/metadata/distribution_manifest.json

@@ -1,6 +1,6 @@
 {
   "schema_version": 1,
-  "version": "5.10.0",
+  "version": "5.11.0",
   "owned_skills": [
     "academic-aio",
     "add-journal",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "medsci-skills",
-  "version": "5.10.0",
+  "version": "5.11.0",
   "description": "MedSci Skills — a medical/scientific research skill suite for AI coding agents (Claude Code, Codex, Cursor, Copilot). The npm package is a terminal-friendly installer shortcut; the canonical distribution remains the GitHub repository and the Claude Code plugin marketplace.",
   "license": "SEE LICENSE IN LICENSE",
   "homepage": "https://github.com/Aperivue/medsci-skills#readme",

package/skills/humanize/SKILL.md

@@ -178,6 +178,7 @@ If the density remains above 2.0, run another fix-verify cycle (max 3 rounds).
 | 19 | § (section sign) marker | "as in §2.3", "(see §Discussion)", "§Results" | Delete or replace with section name ("Methods", "Results") — `grep -c "§"` = 0 |
 | 20 | Methods/Results self-reference parenthetical | "(Methods §X)", "(Results §3.1)", "(Methods, Section 2.3)" | Drop the parenthetical or shorten to "(see Methods)" |
 | 21 | AI Disclosure boilerplate (body) | "## Artificial Intelligence Disclosure", "Generative AI was not used to create..." in manuscript body | Remove from body → place in cover letter / submission form only (per `~/.claude/rules/journal-ai-image-policies.md`) |
+| 25 | Inline-emphasis over-use (typographic over-signposting) | Single-word italics (*into*, *passive*, *same*), whole-clause italics (*a redesign of the relationship itself*), bold used mid-paragraph to signpost | Remove inline emphasis; keep only legitimate italics — statistical symbols (*P*, *t*, *n*), Latin (*in vivo*, *et al.*), gene/species (*BRCA1*). A bold **run-in subheading** at line start is fine (Nature/npj style) |
 
 ### Response-Letter Patterns (R2R)
 
@@ -247,6 +248,7 @@ the pass/fail status.
 | Pattern 19 — `§` symbol | ENFORCED (senior MA reviewer prep) | `grep -c "§" manuscript.md` > 0 | auto-strip; verify post-rewrite count == 0 |
 | Pattern 20 — `(see Methods §X)` self-reference | ENFORCED | match found | rewrite to direct section name reference |
 | Pattern 21 — AI Disclosure paragraph in body | ENFORCED | "Generative AI was not used..." paragraph in manuscript body | move to cover letter or remove |
+| Pattern 25 — inline-emphasis over-use | ENFORCED | italic-emphasis density over threshold after allowlist | run `/self-review` `scripts/check_emphasis_density.py --manuscript manuscript.md`; `EMPHASIS_OVERUSE` (Minor) means strip inline italics (keep only stat symbols / Latin / gene-species); whole-clause italics are the strongest tell |
 | Patterns 22-24 — R2R editing-mechanism / draft line-number / tooling leak | TRIAGE (response letters); `§` = 0 hard | detection greps in ai_patterns.md R2R section surface candidates | review each hit (analysis narration, quoted additions, revised-manuscript page/line are NOT tells); rewrite confirmed tells to substantive prose |
 | Citation preservation invariant | ENFORCED | any pre-existing `[@bibkey]` removed by rewrite | revert that single rewrite; flag for user |
 | Numerical preservation invariant | ENFORCED | any number changed by rewrite | revert; flag for user |

package/skills/peer-review/SKILL.md

@@ -135,11 +135,11 @@ confidential note and the recommendation are consistent.
 
 ### Phase 2A: Systematic Review / Meta-Analysis Extension
 
-Apply this internal-consistency-first gate (P0) plus 17-probe checklist (P1–P17) **only when manuscript type is "Systematic Review", "Meta-Analysis", or "Systematic Review and Meta-Analysis"**. These probes complement (do not replace) the generic Phase 2 issue checklist.
+Apply this internal-consistency-first gate (P0) plus 19-probe checklist (P1–P19) **only when manuscript type is "Systematic Review", "Meta-Analysis", or "Systematic Review and Meta-Analysis"**. These probes complement (do not replace) the generic Phase 2 issue checklist.
 
 **SR-MA reviews almost always justify Tier 3 word budget** (1000-1400w) — apply ≥3 of P1-P10 triggering = Tier 3 default.
 
-**Probe detail (P0–P17), with output templates and the leads-vs-findings discipline:** `${CLAUDE_SKILL_DIR}/references/domain-probes/sr_ma.md`. Load it and apply each probe when the trigger above fires. In this skill, map each probe finding to the review draft as a Major / Minor comment; route conclusion-threatening or integrity findings into the Confidential Comments to the Editor, and place a confirmed error that drives a headline claim as the Major #1 candidate.
+**Probe detail (P0–P19), with output templates and the leads-vs-findings discipline:** `${CLAUDE_SKILL_DIR}/references/domain-probes/sr_ma.md`. Load it and apply each probe when the trigger above fires. In this skill, map each probe finding to the review draft as a Major / Minor comment; route conclusion-threatening or integrity findings into the Confidential Comments to the Editor, and place a confirmed error that drives a headline claim as the Major #1 candidate.
 
 ### Phase 2B: Survival / Prognostic Model Extension
 
@@ -182,14 +182,14 @@ The original-research probes (Phase 2 issue checklist, Phase 2A/2B/2C) do not tr
 
 ### Phase 2E: Observational / Confounding Extension
 
-Apply this 16-probe checklist (O1–O16) **only when the manuscript is an observational study** (cohort, case-control, cross-sectional, health-screening / registry) **whose central claim is an adjusted exposure–outcome association** estimated by covariate adjustment rather than randomization. These probes complement (do not replace) the generic Phase 2 issue checklist and the STROBE reporting items; they target the gap between the stated adjustment set and what the exposure-stratified Table 1 shows.
+Apply this 18-probe checklist (O1–O18) **only when the manuscript is an observational study** (cohort, case-control, cross-sectional, health-screening / registry) **whose central claim is an adjusted exposure–outcome association** estimated by covariate adjustment rather than randomization. These probes complement (do not replace) the generic Phase 2 issue checklist and the STROBE reporting items; they target the gap between the stated adjustment set and what the exposure-stratified Table 1 shows.
 
 **Exempt**:
 - Randomized trials (confounding controlled by design → Phase 2 + CONSORT)
 - Purely descriptive / prevalence reports with no adjusted association claim
 - Diagnostic-accuracy studies with no exposure–outcome estimand (→ Phase 2A DTA cells + categories A–C)
 
-**Probe detail (O1–O16), with output templates:** `${CLAUDE_SKILL_DIR}/references/domain-probes/observational_confounding.md`. Load it and apply each probe when the trigger above fires. O1 (a measured covariate imbalanced by exposure in Table 1 yet absent from the adjustment set), O7 (an outcome consequence/mediator wrongly adjusted — the opposite-direction failure, e.g. serum uric acid in an eGFR model), and O8 (records > subjects with the analysis unit undisclosed) are data-checkable and the highest-yield probes — verify O1/O7 against the manuscript's own Table 1 and run the records-vs-subjects check for O8. In this skill, map each probe finding to the review draft as a Major / Minor comment; a confounding-completeness gap (O1), over-adjustment that moves the headline estimate (O7), a selection/collider structure (O3), undisclosed repeat-subject clustering (O8), an undisclosed complete-case collapse (O5), a report-derived outcome with no construct-validity defence (O9), an inferential effect-size gradient across overlapping/nested subsets with no difference/interaction test (O10), an ignored/mis-specified complex-survey design (O11, NHANES/KNHANES weights without strata+PSU, or a subgroup by row-deletion), a data-mined inflection-point/'saturation' cutoff (O12), a cross-sectional mediation claimed as a causal chain without a temporal-order caveat / M–Y-confounding sensitivity (O13), or a synergy/joint-effect claim on the wrong interaction scale — multiplicative-only or joint-category ORs with no additive RERI/AP/S (O14) — is design-level, so surface it in the Confidential Comments to the Editor and place it as the Major #1 candidate rather than softening it to a reporting fix.
+**Probe detail (O1–O18), with output templates:** `${CLAUDE_SKILL_DIR}/references/domain-probes/observational_confounding.md`. Load it and apply each probe when the trigger above fires. O1 (a measured covariate imbalanced by exposure in Table 1 yet absent from the adjustment set), O7 (an outcome consequence/mediator wrongly adjusted — the opposite-direction failure, e.g. serum uric acid in an eGFR model), and O8 (records > subjects with the analysis unit undisclosed) are data-checkable and the highest-yield probes — verify O1/O7 against the manuscript's own Table 1 and run the records-vs-subjects check for O8. In this skill, map each probe finding to the review draft as a Major / Minor comment; a confounding-completeness gap (O1), over-adjustment that moves the headline estimate (O7), a selection/collider structure (O3), undisclosed repeat-subject clustering (O8), an undisclosed complete-case collapse (O5), a report-derived outcome with no construct-validity defence (O9), an inferential effect-size gradient across overlapping/nested subsets with no difference/interaction test (O10), an ignored/mis-specified complex-survey design (O11, NHANES/KNHANES weights without strata+PSU, or a subgroup by row-deletion), a data-mined inflection-point/'saturation' cutoff (O12), a cross-sectional mediation claimed as a causal chain without a temporal-order caveat / M–Y-confounding sensitivity (O13), or a synergy/joint-effect claim on the wrong interaction scale — multiplicative-only or joint-category ORs with no additive RERI/AP/S (O14) — is design-level, so surface it in the Confidential Comments to the Editor and place it as the Major #1 candidate rather than softening it to a reporting fix.
 
 ### Phase 2E-2: Clinical Prediction-Model Extension
 
@@ -201,7 +201,7 @@ Apply this 4-probe checklist (CP1–CP4) **only when the manuscript develops or
 
 Apply when an AI/ML **primary study** (diagnostic, prognostic, triage, detection) makes a clinical claim in the Title/Abstract/Conclusion — generalizable, outperforms clinicians, deployment-ready, can replace a reader. Complements Phase 2F (recommendation calibration) and the signature "Overclaiming vs evidence level" check; co-applies with Phase 2C for radiomics-AI and Phase 2B for prognostic-AI.
 
-**Probe detail (AO0–AO6), with output templates and the leads-vs-findings discipline:** `${CLAUDE_SKILL_DIR}/references/domain-probes/ai_overclaiming.md`. Load it and apply each probe when the trigger fires. Run AO0 first — locate the load-bearing claim and read it together with its cited evidence before alleging over-reach (a hedged Discussion qualifier is not a headline). In this skill, map each probe finding to the review draft as a Major / Minor comment; a headline generalizability (AO1), superiority/replacement (AO2/AO3), or deployment-readiness (AO4) claim that outruns the design is framing-level — surface it in the Confidential Comments to the Editor and place it as the Major #1 candidate when it is the paper's headline. AO5 catches over-reach in the reported metric itself (best-fold headline without cross-fold CI/SD, unstated/test-tuned operating point, rebalanced-accuracy, or a code-vs-claims mismatch); pair it with the `exemplar_reviews/optimistic_validation_reporting.md` phrasing model and raise it as Major when it carries the headline.
+**Probe detail (AO0–AO7), with output templates and the leads-vs-findings discipline:** `${CLAUDE_SKILL_DIR}/references/domain-probes/ai_overclaiming.md`. Load it and apply each probe when the trigger fires. Run AO0 first — locate the load-bearing claim and read it together with its cited evidence before alleging over-reach (a hedged Discussion qualifier is not a headline). In this skill, map each probe finding to the review draft as a Major / Minor comment; a headline generalizability (AO1), superiority/replacement (AO2/AO3), or deployment-readiness (AO4) claim that outruns the design is framing-level — surface it in the Confidential Comments to the Editor and place it as the Major #1 candidate when it is the paper's headline. AO5 catches over-reach in the reported metric itself (best-fold headline without cross-fold CI/SD, unstated/test-tuned operating point, rebalanced-accuracy, or a code-vs-claims mismatch); pair it with the `exemplar_reviews/optimistic_validation_reporting.md` phrasing model and raise it as Major when it carries the headline.
 
 ### Phase 2G-2: Model-Development / Validation-Integrity Extension
 
@@ -223,9 +223,9 @@ Apply this 8-probe checklist (RC0–RC7) **only when the manuscript is a randomi
 
 ### Phase 2I: Diagnostic-Accuracy / Reader-Study Extension
 
-Apply this 8-probe checklist (D1–D8) **only when the manuscript is a diagnostic test accuracy (DTA) primary study** — an index test against a reference standard — including **multi-reader multi-case (MRMC)** reader studies (AI-vs-reader or modality comparison). These probes complement (do not replace) the generic Phase 2 issue checklist and the STARD / QUADAS-2 items; they target verification/spectrum/blinding bias and the MRMC design/variance issues a reader study adds. (For a DTA **meta-analysis**, use Phase 2A / `sr_ma.md`.)
+Apply this 11-probe checklist (D1–D11) **only when the manuscript is a diagnostic test accuracy (DTA) primary study** — an index test against a reference standard — including **multi-reader multi-case (MRMC)** reader studies (AI-vs-reader or modality comparison). These probes complement (do not replace) the generic Phase 2 issue checklist and the STARD / QUADAS-2 items; they target verification/spectrum/blinding bias and the MRMC design/variance issues a reader study adds. (For a DTA **meta-analysis**, use Phase 2A / `sr_ma.md`.)
 
-**Probe detail (D1–D8), with output templates and the leads-vs-findings discipline:** `${CLAUDE_SKILL_DIR}/references/domain-probes/diagnostic_accuracy.md`. Load it and apply each probe when the trigger fires. In this skill, map each probe finding to the review draft as a Major / Minor comment; two-gate (case-control) sampling (D2), verification/incorporation bias (D1), or an MRMC analysis that ignores reader variance (D6) is design/analysis-level — surface it in the Confidential Comments to the Editor and place it as the Major #1 candidate. Pairs the `analyze-stats` `table-types/reader_study.md` table and the `make-figures` `exemplar_plots/mrmc_roc.md` figure; a test-set-tuned operating threshold pairs with `exemplar_reviews/optimistic_validation_reporting.md`.
+**Probe detail (D1–D11), with output templates and the leads-vs-findings discipline:** `${CLAUDE_SKILL_DIR}/references/domain-probes/diagnostic_accuracy.md`. Load it and apply each probe when the trigger fires. In this skill, map each probe finding to the review draft as a Major / Minor comment; two-gate (case-control) sampling (D2), verification/incorporation bias (D1), or an MRMC analysis that ignores reader variance (D6) is design/analysis-level — surface it in the Confidential Comments to the Editor and place it as the Major #1 candidate. Pairs the `analyze-stats` `table-types/reader_study.md` table and the `make-figures` `exemplar_plots/mrmc_roc.md` figure; a test-set-tuned operating threshold pairs with `exemplar_reviews/optimistic_validation_reporting.md`.
 
 ### Phase 2J: Case-Report Extension
 
@@ -396,7 +396,7 @@ After drafting, verify mechanically:
    - ≥50% of Minor requests use hedged forms ("I'd suggest," "could," "would help") rather than imperative ("must," bare "Please [verb]")
    - General Comments names ≥2 specific strengths before listing concerns
    - At most 1 typo/grammar Minor Comment, only if in formal section or systematic
-9. **SR-MA-specific QC** (if Phase 2A applied): Confirm the P0 internal-consistency gate was run before any fabrication claim. For each P1–P17 probe used, verify the corresponding Major comment cites source PMID + source page/table reference + verbatim quote, and that no probe lead was promoted to a finding without source confirmation (leads-vs-findings discipline). Reviews citing extraction errors without source-page reference are not actionable for authors.
+9. **SR-MA-specific QC** (if Phase 2A applied): Confirm the P0 internal-consistency gate was run before any fabrication claim. For each P1–P19 probe used, verify the corresponding Major comment cites source PMID + source page/table reference + verbatim quote, and that no probe lead was promoted to a finding without source confirmation (leads-vs-findings discipline). Reviews citing extraction errors without source-page reference are not actionable for authors.
 10. **Radiomics-reproducibility QC** (if Phase 2C applied): If an acquisition-parameter sweep predicts an outcome from its own grid axes (R1 design-grid circularity) or the substantive result is a cross-domain failure framed as success (R3), confirm the recommendation reflects design-level severity and is not softened to a reporting fix. Where a model × threshold/cohort grid yields a few p < 0.05, confirm the multiplicity / expected-false-positive count is named (R4), not deferred to "statistical review needed."
 11. **Review-article QC** (if Phase 2D applied): Confirm RV1–RV9 are reflected — in particular that novelty/value-add (RV1) is raised for a saturated topic and that gap-filling (RV8) is present, not just error-spotting. Verify SANRA is used as an appraisal aid, not over-enforced as a reporting guideline (no PRISMA demand on a narrative review; only RV3 is SANRA-aligned and phrased as a suggestion). Verify every suggested addition uses "consider adding" phrasing (no "must cite"), is source-confirmed, and that preprints are labeled as preprints (not equated with peer-reviewed guidelines). Confirm Phase 2F was run for the recommendation: when RV1 novelty is a Major in a saturated space with no distinct contribution, the recommendation is escalated toward Reject (the contribution IS the product — weak novelty is unfixable-in-current-form), not defaulted to the revision/Reconsider tier.
 12. **AI/method/review priority QC**: Before a Major Revision (or Reconsider) recommendation, confirm Phase 2F
@@ -466,7 +466,7 @@ For radiomic feature-reproducibility / phantom parameter-sweep / reliability-fil
 
 For Review / narrative / primer / state-of-the-art manuscripts, apply the Phase 2D 9-probe audit (novelty/value-add, scope/aims, evidence-gathering transparency, technical/medical accuracy, taxonomy/synthesis coherence, balance/currency/citation accuracy, load-bearing figures/tables, constructive gap-filling, curated-base circularity) in place of the original-research probes — error-spotting plus proportionate gap-filling, with SANRA used as an appraisal aid only.
 
-For observational studies whose central claim is an adjusted exposure–outcome association, also apply the Phase 2E 16-probe audit (confounding completeness, adjustment-set provenance, selection/collider bias, exposure measurement validity, missing-data / complete-case collapse, residual-confounding E-value, over-adjustment, analysis-unit/clustering, outcome construct validity, overlapping-subset gradient, complex-survey design & weighting, data-driven threshold mining, cross-sectional mediation, interaction scale, selection on modality/procedure availability, serial-imaging lesion-tracking), with O1 (a measured covariate imbalanced by exposure in Table 1 yet absent from the adjustment set) and O7 (an outcome consequence/mediator wrongly adjusted) checked against the manuscript's own Table 1.
+For observational studies whose central claim is an adjusted exposure–outcome association, also apply the Phase 2E 18-probe audit (confounding completeness, adjustment-set provenance, selection/collider bias, exposure measurement validity, missing-data / complete-case collapse, residual-confounding E-value, over-adjustment, analysis-unit/clustering, outcome construct validity, overlapping-subset gradient, complex-survey design & weighting, data-driven threshold mining, cross-sectional mediation, interaction scale, selection on modality/procedure availability, serial-imaging lesion-tracking, many-exposure agnostic-scan multiplicity, pseudoreplication in multi-rater agreement), with O1 (a measured covariate imbalanced by exposure in Table 1 yet absent from the adjustment set) and O7 (an outcome consequence/mediator wrongly adjusted) checked against the manuscript's own Table 1.
 
 For cross-modality image-synthesis manuscripts (MRI→PET / MRI→CT / non-contrast→contrast / low-dose→full-dose) that claim functional/molecular information or a substitute for the unavailable target modality, also apply the Phase 2K 4-probe audit (IS1 determinism/information-ceiling vs a source→label baseline, IS2 target-derived-preprocessing/slice-selection leakage, IS3 global vs lesion-level quantitative agreement, IS4 mechanistic/proxy-signal plausibility); IS2 and IS4 are typically unfixable-in-current-form and govern the recommendation per Phase 2F.
 

package/skills/peer-review/references/domain-probes/ai_overclaiming.md

@@ -49,6 +49,11 @@ A 5-probe checklist (AO1–AO5, with AO0 as a gate) for medical-AI/ML primary st
   - (b) **Success-conditioned selection** — the arm or the analyzed subset is gated on an AI-success condition (cases where the model produced an output, segmentations that "passed", studies the pipeline did not fail on), so the comparison is conditioned on the very thing under test.
 - This is a **design/paradigm-level** defect: the operationalized task, not the prose, is mis-specified, so it cannot be fixed by rewording the claim — escalate **past an ordinary Major** (editors read it as a Reject-grade construct-validity failure; a panel that files it as a fixable Major under-rates it). The fix is a re-designed arm whose task matches the intended deployment workflow and an unconditioned (consecutive / intention-to-diagnose) analysis set.
 
+**AO7 — Comparative "within/comparable-to X variability" claim whose benchmark X was never quantified**:
+- When the Abstract / hypothesis / Conclusion asserts that a result falls **"within inter-expert variability"**, **"comparable to the reference range"**, or **"on par with human readers"**, the named comparator quantity (the inter-expert spread, the human-reader range) must actually be **computed and reported** in Results / a Table. A benchmark claim whose benchmark does not exist in the paper is an unbenchmarked overclaim — the reader cannot check the comparison because the "within X" bound was never measured (e.g. an agreement-envelope was computed but the *volume* inter-expert spread it is compared against was not).
+- Lead: for each "within / comparable-to / on par with `<comparator>` variability|range|performance" claim, verify the `<comparator>` quantity appears in Results or a Table with a number; if absent → flag as unbenchmarked.
+- Severity: MAJOR when the comparative claim is a headline; the fix is to compute and report the benchmark, or to drop the comparative framing. Distinct from AO5 (metric overstates evidence) — here the comparison *target* is simply missing.
+
 ## Decision-impact / early-deployment probes (DECIDE-AI axis, DI1–DI5)
 
 Co-apply when a study claims **clinical utility, deployment, or decision impact** of an AI system, or *is* an early-stage live clinical evaluation. The reporting axis is then DECIDE-AI (early-stage clinical evaluation of AI decision-support); these probes check that a utility/deployment claim rests on real-use evidence, not retrospective accuracy. They sharpen AO4 for the deployment-evaluation case.

package/skills/peer-review/references/domain-probes/diagnostic_accuracy.md

@@ -43,6 +43,20 @@ A checklist for **diagnostic test accuracy (DTA) primary studies** — an index
 - For an **imaging** study, check that **modality-specific safety contraindications** and **device/artifact exclusions** are enumerated where applicable: MR safety (pacemaker/implant, claustrophobia), iodinated/gadolinium **contrast** contraindications (renal function, allergy, pregnancy), and **image-quality/artifact** exclusions (motion, metal artifact, incomplete coverage). Silent omission of these categories in a prospective imaging cohort understates the selected spectrum.
 - Severity: a flow-vs-prose exclusion mismatch is MAJOR when it changes the analytic-N or the eligible spectrum; missing modality-safety/artifact exclusion categories is MINOR–MAJOR depending on how much of the source population they remove. The fix is a reconciled exclusion list (flow == prose) plus an explicit modality-safety/artifact exclusion enumeration.
 
+**D9 — Confidence-weighted / rating reader study: unweighted baseline AUC must accompany the weighted primary**:
+- When a reader study's novelty is a **confidence-weighted** (or rating-collapsed) score used as the ROC/AUC predictor, the **unweighted binary-call AUC** must be reported side-by-side (as a sensitivity analysis). Without it, you cannot tell whether the weighting *created* the result or hid an estimator fragility (e.g. a folded/non-monotonic encoding that collapses `real/5` with `ai/1`).
+- Lead: if the primary predictor is a confidence/rating→single-score collapse and no unweighted binary-call AUC appears, ask for it; also confirm the (call × confidence) encoding is strictly monotonic (no boundary collision) — the folded-score bug understates one hypothesis and can flip another.
+- Severity: MAJOR when the weighted score is the headline and no unweighted baseline is shown; the weighting must "earn its place" against the simpler estimator.
+
+**D10 — "No stratum met threshold X" vs a per-stratum table that does meet X**:
+- When the manuscript states a numeric admissibility/deployability rule (e.g. "AUC ≥ 0.75 **and** lower 95% bound ≥ 0.70") and concludes "**no stratum met** the rule" / "all strata were below", cross-check that claim against the **per-stratum AUC + CI table**. A blanket negative-stratum claim contradicted by a tabled stratum that literally satisfies the rule (e.g. ultrasound 0.789, CI 0.742–0.834) is a self-contradiction a reviewer verifies with arithmetic.
+- Lead: parse the stated threshold rule, then test each stratum row's estimate+CI against it; flag any "no stratum met" prose contradicted by a qualifying row. Note multiplicity (with k strata, some crossings are expected) — the fix is to report the qualifying strata and frame them as hypothesis-generating, not to deny them.
+- Severity: MAJOR (a headline negative claim contradicted by the paper's own table).
+
+**D11 — Mixed normalisation in one comparison column (apples-to-oranges head-to-head)**:
+- When a table column (or a row-wise "A vs B" comparison) places two metrics computed under **different normalisations / definitions** side by side — e.g. one method's volume error as a *standard relative error* and the comparator's as a *range-normalised* error — the implied head-to-head is invalid; the numbers are not on the same scale.
+- Lead: check whether the two values sharing a comparison column are defined identically in the Methods/footnotes; if not, require either a common-scale recomputation or an explicit "not on the same scale" footnote. Severity: MEDIUM–MAJOR when a superiority/comparability claim rests on the mixed-scale comparison.
+
 **Output template (D2 / D6 example)**:
 > "The study uses a case-control (two-gate) design — confirmed cases versus healthy controls — rather than a consecutive series of patients in whom the diagnosis was suspected. This typically overestimates accuracy and does not reflect the intended-use spectrum, so I'd read the reported sensitivity/specificity as proof-of-concept rather than clinical accuracy, and suggest tempering the Abstract accordingly. Separately, the reader study reports a single reader-averaged AUC; because readers are a sample, I'd suggest an MRMC analysis (e.g., Obuchowski–Rockette) that accounts for both reader and case variance, with per-reader estimates shown and the unit of analysis (per-patient vs per-lesion) stated."
 

package/skills/peer-review/references/domain-probes/observational_confounding.md

@@ -108,6 +108,11 @@ A 17-probe checklist for observational studies (cohort, case-control, cross-sect
 - **(b) Independent replication is the real safeguard, not the correction alone.** Agnostic scans carry a high false-discovery proportion even after FDR; the load-bearing control is an **independent replication** — a held-out split, a different cohort, or a different survey cycle — requiring **directional concordance** and a reported **replication rate** (how many discovery hits replicated, e.g. "110/164 replicated", "59% Bonferroni-surviving and concordant"). A single-cohort FDR-significant scan with no replication is hypothesis-generating only; presenting its top hits as established → MAJOR (downgrade to exploratory or add replication). A held-out **validation** set must be used once, for the final model, not reused as a second discovery pass.
 - **(c) Correlated exposures — Bonferroni conservatism and co-exposure confounding.** Exposome/omics exposures are heavily intercorrelated, with two consequences: (i) Bonferroni on the raw count is **over-conservative** (the effective number of independent tests is smaller — a permutation threshold or an effective-number adjustment is more honest), and (ii) a **univariate** hit can be a *marker for a correlated true cause* rather than itself causal, because the single-exposure model ignores co-exposure confounding / mixtures. Is the correlation structure addressed (effective-number-of-tests, clustering / dimension reduction, or a multi-exposure / mixture model), and is a univariate hit read as a *screen* rather than an isolated causal effect? A causal/actionable claim from one univariate hit inside a correlated exposure block, with no co-exposure adjustment → MAJOR.
 - **(d) Selective top-hit reporting / hidden multiple testing (HARKing).** The **full tested-exposure set and the full results** (every effect size + p/q, not just the significant rows) must be available — ideally pre-registered, at minimum in a supplement. Reporting only the significant hits, or re-narrating the scan as a "focused a-priori hypothesis" about the winners after seeing the results, is selective reporting / hidden multiplicity. Nominal (uncorrected) hits may be shown to inform future work only if **explicitly labelled as not surviving correction**. A Results section that presents only winners, with no full results table and no pre-registration → MAJOR (reporting).
+
+**O18 — Pseudoreplication in multi-rater agreement / reader studies (pooled pairwise vs per-subject)**:
+- When an agreement or reader study computes a test (Mann–Whitney, t-test, correlation) on **pooled pairwise distances / reader-pairs** rather than on **independent units** (subjects, lesions, aneurysms), the effective n is inflated (each subject contributes several pairwise rows) and the p-value is anti-conservative. This is the reader/agreement sibling of the analysis-unit / clustering issue in O8: the number of *observations* is not the number of *independent units*.
+- Lead: when a reported test **n exceeds the number of independent subjects/lesions** (e.g. n=448 or 672 pairwise from 112 aneurysms) and no clustering / mixed-effects / per-subject aggregation is stated → flag. Ask for the analysis re-run at the **per-subject** level (or a mixed model with a subject random effect); a pooled-pairwise p can flip (e.g. p=0.02 pooled → p=0.59 per-aneurysm).
+- Severity: MAJOR when a headline agreement/superiority claim rests on the pooled-pairwise test; MINOR when a per-subject sensitivity reproduces it. Applies to all multi-rater agreement / MRMC reader studies.
 - **(e) Effect size and resolution honesty at scale.** With the very large N these scans run on, trivially small effects clear any threshold; report **effect magnitudes** and clinical relevance alongside significance, and respect the **resolution floor** (a permutation procedure with k permutations cannot resolve p below ≈ 1/k; FDR has a minimum detectable q at a given hit count). A "number of significant exposures" headline with no effect sizes overstates the finding.
 - Severity: MAJOR when a headline causal/actionable claim rests on uncorrected, single-cohort-only, or univariate-in-a-correlated-block top hits; MINOR when correction + replication + full results are present and the claim is framed as screening. Cross-link O11 (complex-survey design — an NHANES/KNHANES ExWAS must combine design-based standard errors with the multiplicity correction, not one or the other), O12 (the single-exposure threshold/non-linearity analogue), and O2/O7 (confounding / over-adjustment for whichever hit survives). Report the tested-set size, the correction method, and the replication design explicitly.
 

package/skills/peer-review/references/domain-probes/sr_ma.md

@@ -110,6 +110,17 @@ Internal-consistency-first gate (P0) plus an 11-probe checklist (P1–P11). Thes
 - Lead: count studies at the 0%/100% boundary per outcome; if ≥ (k−1) are at the boundary, flag the pooled point-estimate/CI as spurious precision.
 - Recommend a descriptive tally ("5 of 6 studies reported 100%; the sixth reported X%") rather than a pooled estimate. Severity: MAJOR when the boundary pool is presented as a headline accuracy/success figure.
 
+**P18 — Train-vs-validation pool integrity: apparent estimate smuggled into the "validation" pool**:
+- A DTA / prognostic-model MA that separates a "training" pool from a "validation" pool must build the validation pool from **held-out / external / cross-validated** estimates only. The failure mode: for studies without cross-validation, the authors drop each study's **apparent (in-sample / resubstitution)** sensitivity–specificity (or "overall estimate") INTO the validation pool — so the headline "validation" performance is partly the same in-sample numbers, and the "train ≈ validation stability" the Discussion leans on is mechanical reuse.
+- Lead: grep Methods for a train/validation split co-occurring with {`overall estimate`, `non–cross-validated` / `non-cross-validated`, `apparent`, `resubstitution`, `where validation was unavailable … used the overall`}. Then check whether a **sensitivity analysis excluding non-CV studies materially moves the validation pool** (a large drop = the primary "validation" estimate is optimism-contaminated).
+- Severity: MAJOR #1 when the headline rests on it (estimand incoherence — the "validation" pool is not a validation estimate); often unfixable in the current form (recompute restricted to genuinely held-out estimates, or relabel as apparent). See `~/.claude/rules/dta-meta-analysis.md`.
+
+**P19 — Reviewer-side included-study cell audit: metric-type, self-eligibility contradiction, CI provenance**:
+- As a reviewer, a **random/convenience sample of the open-access included studies** cell-checked against source is sufficient to demonstrate a systematic extraction-error class — full re-extraction is NOT the reviewer's job. A sampled error rate above ~30% ⇒ the pooled estimates are untrustworthy without author-side full cell reconciliation (P1 / `dta-meta-analysis.md` §1).
+- Beyond value-matches-source (P1), check three extraction pathologies: **(a) metric-type identity** — an `accuracy` or a *different task's* AUC tabulated AS the target AUC (accuracy ↔ AUC ↔ sens/spec are not interchangeable); **(b) self-eligibility contradiction** — an extracted metric/design that violates the review's OWN stated inclusion rule (e.g. an accuracy-only study the eligibility text excludes, entered anyway); **(c) CI provenance** — a CI present in the MA table that **does not exist in the source** ("derived from the point estimate and sample size" is not a valid route to an AUC CI).
+- Also flag validation-type / analysis-unit mislabels (5-fold CV recorded as train/validation; N scans counted as N patients). Absence of a per-study 2×2 / sens–spec table is itself a PRISMA-DTA reporting MAJOR.
+- Lead: sample K OA included studies; per study confirm metric TYPE + value + that the extracted design satisfies the paper's own eligibility + that any tabled CI appears in the source. Severity scales with the sampled error rate.
+
 **Output template (P1 cell-swap example)**:
 > "I spot-checked [Author Year] (PMID [...]) against the source paper and found that the values in Figure X are swapped. The source paper reports external-test sensitivity A% / specificity B% (n=N); the manuscript forest entries place [num1/denom1] in the sensitivity slot (which is the source's specificity numerator/denominator) and [num2/denom2] in the specificity slot (which is the source's sensitivity)."
 

package/skills/self-review/SKILL.md

@@ -329,7 +329,7 @@ publication bias (funnel plot, Egger), and sensitivity/subgroup analyses.
 
 **Type-Specific Additional Checks:**
 
-- **Observational studies**: Confounding assessment (DAG or adjustment strategy), selection bias, exposure measurement validity. Run **Phase 2.5e (Confounding Completeness)** and apply the O1–O16 probes in `references/domain-probes/observational_confounding.md` — including O7 (over-adjustment: do not adjust for a consequence/mediator of the outcome, e.g. serum uric acid in an eGFR model — the opposite-direction failure to O1), O8 (analysis unit & clustering — run `check_cohort_arithmetic.py --id-col` for records-vs-subjects), O9 (construct validity of a report-/registry-derived outcome), O10 (an inferential effect-size gradient across overlapping/nested subsets needs a difference/interaction test, not descriptive refinement alone), and — for complex-survey data (NHANES/KNHANES/CHNS) — O11 (design-based weighting: the right weight + strata + PSU, subpopulation-not-subset) and O12 (data-driven inflection-point/'saturation' threshold mining needs a breakpoint CI + pre-specification, not a quoted cutoff), O13 (a cross-sectional mediation claim cannot establish X→M→Y order and needs an unmeasured-M–Y-confounding sensitivity), and O14 (a synergy/joint-effect/effect-modification claim needs the additive scale — RERI/AP/S with CIs — not a multiplicative-only interaction or joint-category ORs), O15 (an analytic cohort selected on an optional modality/procedure's availability is a spectrum/selection bias, not a generalizability caveat — ask for consecutive enrollment + a selected-vs-source comparison), and O16 (a serial-imaging size/growth endpoint needs a stated lesion-tracking rule + multiplicity prevalence + a solitary-lesion sensitivity). If the manuscript develops or compares a **clinical prediction model** (TRIPOD / TRIPOD+AI, nested predictor-set comparison), also apply the CP1–CP4 probes in `references/domain-probes/clinical_prediction_model.md` (apparent-vs-optimism-corrected calibration/DCA, the incremental-value-vs-marginal-effect two-null distinction, EPV per nested model, net benefit as model comparison not policy).
+- **Observational studies**: Confounding assessment (DAG or adjustment strategy), selection bias, exposure measurement validity. Run **Phase 2.5e (Confounding Completeness)** and apply the O1–O18 probes in `references/domain-probes/observational_confounding.md` — including O7 (over-adjustment: do not adjust for a consequence/mediator of the outcome, e.g. serum uric acid in an eGFR model — the opposite-direction failure to O1), O8 (analysis unit & clustering — run `check_cohort_arithmetic.py --id-col` for records-vs-subjects), O9 (construct validity of a report-/registry-derived outcome), O10 (an inferential effect-size gradient across overlapping/nested subsets needs a difference/interaction test, not descriptive refinement alone), and — for complex-survey data (NHANES/KNHANES/CHNS) — O11 (design-based weighting: the right weight + strata + PSU, subpopulation-not-subset) and O12 (data-driven inflection-point/'saturation' threshold mining needs a breakpoint CI + pre-specification, not a quoted cutoff), O13 (a cross-sectional mediation claim cannot establish X→M→Y order and needs an unmeasured-M–Y-confounding sensitivity), and O14 (a synergy/joint-effect/effect-modification claim needs the additive scale — RERI/AP/S with CIs — not a multiplicative-only interaction or joint-category ORs), O15 (an analytic cohort selected on an optional modality/procedure's availability is a spectrum/selection bias, not a generalizability caveat — ask for consecutive enrollment + a selected-vs-source comparison), and O16 (a serial-imaging size/growth endpoint needs a stated lesion-tracking rule + multiplicity prevalence + a solitary-lesion sensitivity), O17 (a many-exposure agnostic scan — ExWAS/EWAS/MWAS — needs multiplicity control against the true denominator + independent replication, not a raw p<0.05 top hit), and O18 (a multi-rater agreement / reader test run on pooled pairwise distances rather than independent subjects is pseudoreplication — re-run per-subject). If the manuscript develops or compares a **clinical prediction model** (TRIPOD / TRIPOD+AI, nested predictor-set comparison), also apply the CP1–CP4 probes in `references/domain-probes/clinical_prediction_model.md` (apparent-vs-optimism-corrected calibration/DCA, the incremental-value-vs-marginal-effect two-null distinction, EPV per nested model, net benefit as model comparison not policy).
 - **Educational studies**: Learning outcome measurement validity, Kirkpatrick level, control group adequacy, curriculum fidelity
 - **Meta-analyses**: Search comprehensiveness (2+ databases), screening reproducibility (2 reviewers), RoB assessment per study, GRADE certainty
 - **Case reports**: Diagnostic reasoning transparency, timeline completeness, informed consent, generalizability disclaimer
@@ -341,16 +341,16 @@ These modules carry the same domain-specific critique probes used by `/peer-revi
 
 | Manuscript type / signal | Probe module |
 |---|---|
-| Systematic Review / Meta-Analysis | `references/domain-probes/sr_ma.md` (P0–P17) |
+| Systematic Review / Meta-Analysis | `references/domain-probes/sr_ma.md` (P0–P19) |
 | Time-to-event / survival / prognostic model (Cox, Fine-Gray, DeepSurv, nomogram, risk-stratification cutoff) | `references/domain-probes/survival_prognostic.md` (S1–S9) |
 | Radiomic feature reproducibility / acquisition-parameter sweep / reliability-based feature filtering | `references/domain-probes/radiomics.md` (R1–R4) |
 | Cross-modality image synthesis (MRI→PET / MRI→CT / non-contrast→contrast / low-dose→full-dose) claiming functional/molecular information or target-modality substitution | `references/domain-probes/image_synthesis.md` (IS1–IS4) |
 | Narrative / review article / primer / state-of-the-art | `references/domain-probes/narrative_review.md` (RV1–RV9) |
-| AI/ML primary study with a clinical claim (generalizable / outperforms clinicians / deployment-ready / can replace a reader) | `references/domain-probes/ai_overclaiming.md` (AO0–AO6) |
+| AI/ML primary study with a clinical claim (generalizable / outperforms clinicians / deployment-ready / can replace a reader) | `references/domain-probes/ai_overclaiming.md` (AO0–AO7) |
 | Engineer-built medical-imaging model (segmentation / classification / detection; CNN / U-Net / nnU-Net / transformer) being validated — partition/leakage, seed & run variance, metric selection, reproducibility, reference-standard quality | `references/domain-probes/model_development.md` (MD0–MD8) |
 | LLM / MLLM evaluated on a clinical task (radiology report generation, visual question answering, clinical text extraction/classification; closed API or open weights) | `references/domain-probes/mllm_evaluation.md` (ME0–ME8) |
 | Randomised controlled trial (parallel / crossover / cluster / stepped-wedge) | `references/domain-probes/rct_trial.md` (RC0–RC7) |
-| Diagnostic test accuracy (DTA) primary study / multi-reader multi-case (MRMC) reader study (index test vs reference standard, AI-vs-reader, modality comparison) | `references/domain-probes/diagnostic_accuracy.md` (D1–D8) |
+| Diagnostic test accuracy (DTA) primary study / multi-reader multi-case (MRMC) reader study (index test vs reference standard, AI-vs-reader, modality comparison) | `references/domain-probes/diagnostic_accuracy.md` (D1–D11) |
 | Case report / case series / single-patient clinical narrative (incl. adverse-event/pharmacovigilance and imaging-led radiology/nuclear-medicine/IR reports) | `references/domain-probes/case_report.md` (CR1–CR9) |
 | AI/ML, prediction, or diagnostic study claiming cross-population performance (generalizable / deployment-ready / "works for patients"), or presenting subgroup analyses as a fairness/equity argument | `references/domain-probes/equity_fairness.md` (EQ0–EQ6) |
 | Mendelian randomization (genetic variants as instrumental variables: two-sample summary-data, one-sample, multivariable MR, drug-target / cis-MR, non-linear MR) | `references/domain-probes/mendelian_randomization.md` (MR1–MR8) |
@@ -392,6 +392,20 @@ Internal consistency (Phase 2.5) is necessary but not sufficient. Numbers can be
 consistent across Abstract / Table / Text and still be wrong at the source — a single
 transcription error propagates cleanly through every downstream stage.
 
+Also run the **displayed-arithmetic** gate — a stated difference must equal the subtraction of
+its two displayed component values at the SAME precision:
+
+```bash
+python3 "${CLAUDE_SKILL_DIR}/scripts/check_rounded_delta.py" \
+  --manuscript manuscript.md --out qc/rounded_delta.json
+```
+
+`ROUNDED_DELTA_MISMATCH` (Minor) fires when e.g. AUCs are shown as `0.70` and `0.73` (a displayed
+gap of 0.03) while the between-arm difference is stated as `0.02` — self-consistent only on the
+unrounded values. Fix: report components and the delta at one precision, or footnote that the delta
+is computed on unrounded values. A higher-precision component pair (`0.703` vs `0.726`) with a 2-dp
+delta is the legitimate unrounded case and is not flagged.
+
 **Precedent failure pattern:**
 > A revision-era comparative meta-analysis reported a safety-outcome 2x2 with the
 > arm-level events direction-reversed relative to the primary-source Table. Internal
@@ -743,6 +757,18 @@ Phase 2.5c covers reference **integrity** — are the cited references real (fab
 
 ### Phase 2.5d: Cross-Reference QC (Manuscript ↔ rendered DOCX)
 
+Before the DOCX is built, run the **markdown-stage orphan gate** — every captioned
+`Figure N.` / `Table N.` must be cited at least once elsewhere in the body:
+
+```bash
+python3 "${CLAUDE_SKILL_DIR}/scripts/check_figure_citation.py" \
+  --manuscript manuscript.md --out qc/figure_citation.json
+```
+
+`FIGURE_ORPHAN` / `TABLE_ORPHAN` (Minor) catch a newly-added float that has a legend
+but no in-text citation — the early, no-build counterpart to `check_xref`'s `UNCITED`
+verdict, which catches the same class on the rendered DOCX (below).
+
 Reference-list integrity (Phase 2.5c) does **not** cover Table/Figure
 cross-references. This is a separate failure mode where in-text citations
 ("Supplementary Table S4 reports a sensitivity analysis") resolve to a different
@@ -1081,7 +1107,7 @@ The panel simulates independent peer reviewers who do not see each other's comme
 | Survival / prognostic cohort | R1 Biostatistics & Study Design · R2 Clinical (domain) · R3 Imaging/Radiology (if an imaging exposure) | `references/domain-probes/survival_prognostic.md` |
 | Systematic review / meta-analysis | R1 Methodology (search/screening/PRISMA) · R2 Clinical · R3 Statistics (pooling/heterogeneity) | `references/domain-probes/sr_ma.md` |
 | Radiomics / feature reproducibility | R1 Imaging physics & acquisition · R2 ML / Statistics · R3 Clinical translation | `references/domain-probes/radiomics.md` |
-| Diagnostic-accuracy / AI model | R1 Study design & leakage · R2 Statistics (DeLong, calibration) · R3 Clinical / reference standard | `references/domain-probes/sr_ma.md` (P1 DTA cells) + `references/domain-probes/ai_overclaiming.md` (AO0–AO6, for AI clinical claims) + categories A–C |
+| Diagnostic-accuracy / AI model | R1 Study design & leakage · R2 Statistics (DeLong, calibration) · R3 Clinical / reference standard | `references/domain-probes/sr_ma.md` (P1 DTA cells) + `references/domain-probes/ai_overclaiming.md` (AO0–AO7, for AI clinical claims) + categories A–C |
 | Observational (STROBE) | R1 Epidemiology / confounding · R2 Clinical · R3 Statistics | `references/domain-probes/observational_confounding.md` (O1/O8 run as the Phase 2.5e / `check_cohort_arithmetic.py --id-col` deterministic gates; O7 over-adjustment) + `references/domain-probes/clinical_prediction_model.md` (CP1–CP4, when it is a prediction-model paper) + categories A–J + the effect-size / added-value axes |
 | Narrative / review article | R1 Domain-content expert · R2 Methodology / SANRA · R3 Technical accuracy · R4 Adversarial reject-hunter (structural: RV9 curated-base circularity, RV6 single-anchor overload, RV8 self-citation architecture) | `references/domain-probes/narrative_review.md` |
 | Case report | R1 Clinical case-report reviewer · R2 Ethics / de-identification · R3 Literature-context reviewer | `references/domain-probes/case_report.md` + CARE items + categories D/F/G |

package/skills/self-review/references/domain-probes/ai_overclaiming.md

@@ -49,6 +49,11 @@ A 5-probe checklist (AO1–AO5, with AO0 as a gate) for medical-AI/ML primary st
   - (b) **Success-conditioned selection** — the arm or the analyzed subset is gated on an AI-success condition (cases where the model produced an output, segmentations that "passed", studies the pipeline did not fail on), so the comparison is conditioned on the very thing under test.
 - This is a **design/paradigm-level** defect: the operationalized task, not the prose, is mis-specified, so it cannot be fixed by rewording the claim — escalate **past an ordinary Major** (editors read it as a Reject-grade construct-validity failure; a panel that files it as a fixable Major under-rates it). The fix is a re-designed arm whose task matches the intended deployment workflow and an unconditioned (consecutive / intention-to-diagnose) analysis set.
 
+**AO7 — Comparative "within/comparable-to X variability" claim whose benchmark X was never quantified**:
+- When the Abstract / hypothesis / Conclusion asserts that a result falls **"within inter-expert variability"**, **"comparable to the reference range"**, or **"on par with human readers"**, the named comparator quantity (the inter-expert spread, the human-reader range) must actually be **computed and reported** in Results / a Table. A benchmark claim whose benchmark does not exist in the paper is an unbenchmarked overclaim — the reader cannot check the comparison because the "within X" bound was never measured (e.g. an agreement-envelope was computed but the *volume* inter-expert spread it is compared against was not).
+- Lead: for each "within / comparable-to / on par with `<comparator>` variability|range|performance" claim, verify the `<comparator>` quantity appears in Results or a Table with a number; if absent → flag as unbenchmarked.
+- Severity: MAJOR when the comparative claim is a headline; the fix is to compute and report the benchmark, or to drop the comparative framing. Distinct from AO5 (metric overstates evidence) — here the comparison *target* is simply missing.
+
 ## Decision-impact / early-deployment probes (DECIDE-AI axis, DI1–DI5)
 
 Co-apply when a study claims **clinical utility, deployment, or decision impact** of an AI system, or *is* an early-stage live clinical evaluation. The reporting axis is then DECIDE-AI (early-stage clinical evaluation of AI decision-support); these probes check that a utility/deployment claim rests on real-use evidence, not retrospective accuracy. They sharpen AO4 for the deployment-evaluation case.

package/skills/self-review/references/domain-probes/diagnostic_accuracy.md

@@ -43,6 +43,20 @@ A checklist for **diagnostic test accuracy (DTA) primary studies** — an index
 - For an **imaging** study, check that **modality-specific safety contraindications** and **device/artifact exclusions** are enumerated where applicable: MR safety (pacemaker/implant, claustrophobia), iodinated/gadolinium **contrast** contraindications (renal function, allergy, pregnancy), and **image-quality/artifact** exclusions (motion, metal artifact, incomplete coverage). Silent omission of these categories in a prospective imaging cohort understates the selected spectrum.
 - Severity: a flow-vs-prose exclusion mismatch is MAJOR when it changes the analytic-N or the eligible spectrum; missing modality-safety/artifact exclusion categories is MINOR–MAJOR depending on how much of the source population they remove. The fix is a reconciled exclusion list (flow == prose) plus an explicit modality-safety/artifact exclusion enumeration.
 
+**D9 — Confidence-weighted / rating reader study: unweighted baseline AUC must accompany the weighted primary**:
+- When a reader study's novelty is a **confidence-weighted** (or rating-collapsed) score used as the ROC/AUC predictor, the **unweighted binary-call AUC** must be reported side-by-side (as a sensitivity analysis). Without it, you cannot tell whether the weighting *created* the result or hid an estimator fragility (e.g. a folded/non-monotonic encoding that collapses `real/5` with `ai/1`).
+- Lead: if the primary predictor is a confidence/rating→single-score collapse and no unweighted binary-call AUC appears, ask for it; also confirm the (call × confidence) encoding is strictly monotonic (no boundary collision) — the folded-score bug understates one hypothesis and can flip another.
+- Severity: MAJOR when the weighted score is the headline and no unweighted baseline is shown; the weighting must "earn its place" against the simpler estimator.
+
+**D10 — "No stratum met threshold X" vs a per-stratum table that does meet X**:
+- When the manuscript states a numeric admissibility/deployability rule (e.g. "AUC ≥ 0.75 **and** lower 95% bound ≥ 0.70") and concludes "**no stratum met** the rule" / "all strata were below", cross-check that claim against the **per-stratum AUC + CI table**. A blanket negative-stratum claim contradicted by a tabled stratum that literally satisfies the rule (e.g. ultrasound 0.789, CI 0.742–0.834) is a self-contradiction a reviewer verifies with arithmetic.
+- Lead: parse the stated threshold rule, then test each stratum row's estimate+CI against it; flag any "no stratum met" prose contradicted by a qualifying row. Note multiplicity (with k strata, some crossings are expected) — the fix is to report the qualifying strata and frame them as hypothesis-generating, not to deny them.
+- Severity: MAJOR (a headline negative claim contradicted by the paper's own table).
+
+**D11 — Mixed normalisation in one comparison column (apples-to-oranges head-to-head)**:
+- When a table column (or a row-wise "A vs B" comparison) places two metrics computed under **different normalisations / definitions** side by side — e.g. one method's volume error as a *standard relative error* and the comparator's as a *range-normalised* error — the implied head-to-head is invalid; the numbers are not on the same scale.
+- Lead: check whether the two values sharing a comparison column are defined identically in the Methods/footnotes; if not, require either a common-scale recomputation or an explicit "not on the same scale" footnote. Severity: MEDIUM–MAJOR when a superiority/comparability claim rests on the mixed-scale comparison.
+
 **Output template (D2 / D6 example)**:
 > "The study uses a case-control (two-gate) design — confirmed cases versus healthy controls — rather than a consecutive series of patients in whom the diagnosis was suspected. This typically overestimates accuracy and does not reflect the intended-use spectrum, so I'd read the reported sensitivity/specificity as proof-of-concept rather than clinical accuracy, and suggest tempering the Abstract accordingly. Separately, the reader study reports a single reader-averaged AUC; because readers are a sample, I'd suggest an MRMC analysis (e.g., Obuchowski–Rockette) that accounts for both reader and case variance, with per-reader estimates shown and the unit of analysis (per-patient vs per-lesion) stated."
 

package/skills/self-review/references/domain-probes/observational_confounding.md

@@ -108,6 +108,11 @@ A 17-probe checklist for observational studies (cohort, case-control, cross-sect
 - **(b) Independent replication is the real safeguard, not the correction alone.** Agnostic scans carry a high false-discovery proportion even after FDR; the load-bearing control is an **independent replication** — a held-out split, a different cohort, or a different survey cycle — requiring **directional concordance** and a reported **replication rate** (how many discovery hits replicated, e.g. "110/164 replicated", "59% Bonferroni-surviving and concordant"). A single-cohort FDR-significant scan with no replication is hypothesis-generating only; presenting its top hits as established → MAJOR (downgrade to exploratory or add replication). A held-out **validation** set must be used once, for the final model, not reused as a second discovery pass.
 - **(c) Correlated exposures — Bonferroni conservatism and co-exposure confounding.** Exposome/omics exposures are heavily intercorrelated, with two consequences: (i) Bonferroni on the raw count is **over-conservative** (the effective number of independent tests is smaller — a permutation threshold or an effective-number adjustment is more honest), and (ii) a **univariate** hit can be a *marker for a correlated true cause* rather than itself causal, because the single-exposure model ignores co-exposure confounding / mixtures. Is the correlation structure addressed (effective-number-of-tests, clustering / dimension reduction, or a multi-exposure / mixture model), and is a univariate hit read as a *screen* rather than an isolated causal effect? A causal/actionable claim from one univariate hit inside a correlated exposure block, with no co-exposure adjustment → MAJOR.
 - **(d) Selective top-hit reporting / hidden multiple testing (HARKing).** The **full tested-exposure set and the full results** (every effect size + p/q, not just the significant rows) must be available — ideally pre-registered, at minimum in a supplement. Reporting only the significant hits, or re-narrating the scan as a "focused a-priori hypothesis" about the winners after seeing the results, is selective reporting / hidden multiplicity. Nominal (uncorrected) hits may be shown to inform future work only if **explicitly labelled as not surviving correction**. A Results section that presents only winners, with no full results table and no pre-registration → MAJOR (reporting).
+
+**O18 — Pseudoreplication in multi-rater agreement / reader studies (pooled pairwise vs per-subject)**:
+- When an agreement or reader study computes a test (Mann–Whitney, t-test, correlation) on **pooled pairwise distances / reader-pairs** rather than on **independent units** (subjects, lesions, aneurysms), the effective n is inflated (each subject contributes several pairwise rows) and the p-value is anti-conservative. This is the reader/agreement sibling of the analysis-unit / clustering issue in O8: the number of *observations* is not the number of *independent units*.
+- Lead: when a reported test **n exceeds the number of independent subjects/lesions** (e.g. n=448 or 672 pairwise from 112 aneurysms) and no clustering / mixed-effects / per-subject aggregation is stated → flag. Ask for the analysis re-run at the **per-subject** level (or a mixed model with a subject random effect); a pooled-pairwise p can flip (e.g. p=0.02 pooled → p=0.59 per-aneurysm).
+- Severity: MAJOR when a headline agreement/superiority claim rests on the pooled-pairwise test; MINOR when a per-subject sensitivity reproduces it. Applies to all multi-rater agreement / MRMC reader studies.
 - **(e) Effect size and resolution honesty at scale.** With the very large N these scans run on, trivially small effects clear any threshold; report **effect magnitudes** and clinical relevance alongside significance, and respect the **resolution floor** (a permutation procedure with k permutations cannot resolve p below ≈ 1/k; FDR has a minimum detectable q at a given hit count). A "number of significant exposures" headline with no effect sizes overstates the finding.
 - Severity: MAJOR when a headline causal/actionable claim rests on uncorrected, single-cohort-only, or univariate-in-a-correlated-block top hits; MINOR when correction + replication + full results are present and the claim is framed as screening. Cross-link O11 (complex-survey design — an NHANES/KNHANES ExWAS must combine design-based standard errors with the multiplicity correction, not one or the other), O12 (the single-exposure threshold/non-linearity analogue), and O2/O7 (confounding / over-adjustment for whichever hit survives). Report the tested-set size, the correction method, and the replication design explicitly.
 

package/skills/self-review/references/domain-probes/sr_ma.md

@@ -110,6 +110,17 @@ Internal-consistency-first gate (P0) plus an 11-probe checklist (P1–P11). Thes
 - Lead: count studies at the 0%/100% boundary per outcome; if ≥ (k−1) are at the boundary, flag the pooled point-estimate/CI as spurious precision.
 - Recommend a descriptive tally ("5 of 6 studies reported 100%; the sixth reported X%") rather than a pooled estimate. Severity: MAJOR when the boundary pool is presented as a headline accuracy/success figure.
 
+**P18 — Train-vs-validation pool integrity: apparent estimate smuggled into the "validation" pool**:
+- A DTA / prognostic-model MA that separates a "training" pool from a "validation" pool must build the validation pool from **held-out / external / cross-validated** estimates only. The failure mode: for studies without cross-validation, the authors drop each study's **apparent (in-sample / resubstitution)** sensitivity–specificity (or "overall estimate") INTO the validation pool — so the headline "validation" performance is partly the same in-sample numbers, and the "train ≈ validation stability" the Discussion leans on is mechanical reuse.
+- Lead: grep Methods for a train/validation split co-occurring with {`overall estimate`, `non–cross-validated` / `non-cross-validated`, `apparent`, `resubstitution`, `where validation was unavailable … used the overall`}. Then check whether a **sensitivity analysis excluding non-CV studies materially moves the validation pool** (a large drop = the primary "validation" estimate is optimism-contaminated).
+- Severity: MAJOR #1 when the headline rests on it (estimand incoherence — the "validation" pool is not a validation estimate); often unfixable in the current form (recompute restricted to genuinely held-out estimates, or relabel as apparent). See `~/.claude/rules/dta-meta-analysis.md`.
+
+**P19 — Reviewer-side included-study cell audit: metric-type, self-eligibility contradiction, CI provenance**:
+- As a reviewer, a **random/convenience sample of the open-access included studies** cell-checked against source is sufficient to demonstrate a systematic extraction-error class — full re-extraction is NOT the reviewer's job. A sampled error rate above ~30% ⇒ the pooled estimates are untrustworthy without author-side full cell reconciliation (P1 / `dta-meta-analysis.md` §1).
+- Beyond value-matches-source (P1), check three extraction pathologies: **(a) metric-type identity** — an `accuracy` or a *different task's* AUC tabulated AS the target AUC (accuracy ↔ AUC ↔ sens/spec are not interchangeable); **(b) self-eligibility contradiction** — an extracted metric/design that violates the review's OWN stated inclusion rule (e.g. an accuracy-only study the eligibility text excludes, entered anyway); **(c) CI provenance** — a CI present in the MA table that **does not exist in the source** ("derived from the point estimate and sample size" is not a valid route to an AUC CI).
+- Also flag validation-type / analysis-unit mislabels (5-fold CV recorded as train/validation; N scans counted as N patients). Absence of a per-study 2×2 / sens–spec table is itself a PRISMA-DTA reporting MAJOR.
+- Lead: sample K OA included studies; per study confirm metric TYPE + value + that the extracted design satisfies the paper's own eligibility + that any tabled CI appears in the source. Severity scales with the sampled error rate.
+
 **Output template (P1 cell-swap example)**:
 > "I spot-checked [Author Year] (PMID [...]) against the source paper and found that the values in Figure X are swapped. The source paper reports external-test sensitivity A% / specificity B% (n=N); the manuscript forest entries place [num1/denom1] in the sensitivity slot (which is the source's specificity numerator/denominator) and [num2/denom2] in the specificity slot (which is the source's sensitivity)."