A Risk Framework for AI-Authored Medical Papers

1. Motivation

Medical literature is consumed not only by researchers but also by clinicians making decisions at the bedside, by guideline writers, and increasingly by clinical decision-support systems that ingest published abstracts as training or retrieval sources. The introduction of AI authorship into this stream creates a heterogeneous risk surface that flat disclosure rules under-serve.

A Tier 1 commentary that proposes a research direction is qualitatively different from a Tier 4 meta-analysis whose pooled odds ratio could shift a guideline. Yet existing journal AI policies treat them identically.

We propose RX-RISK, a four-tier framework that:

• Provides a reproducible mapping from manuscript $\to$ tier.
• Specifies tier-conditional disclosure, validation, and review requirements.
• Quantifies the expected reduction in patient-relevant error from adoption.

2. Threat Model

Three harm pathways are central:

1. Hallucinated evidence: invented citations, dose values, or trial outcomes.
2. Plausibility-laundered conclusions: AI-smoothed prose that masks weak underlying inference.
3. Downstream propagation: an AI-authored claim is ingested into a guideline or a CDS retrieval index.

The expected harm $H$ from a manuscript is the product of error probability $p_e$, clinical-decision weight $w$, and reversibility factor $\rho$:

$$H = p_e \cdot w \cdot (1 - \rho).$$

RX-RISK is essentially a tractable proxy for this product.

3. The Four Tiers

Tier 1 — Hypothesis or commentary

Non-quantitative, non-recommendation. Required: basic AI-disclosure tag.

Tier 2 — Descriptive empirical work

Observational reports, case series, secondary analyses without recommendation. Required: AI-disclosure plus the inventory in AI-REPORT-A.

Tier 3 — Inferential / interventional work

RCTs, meta-analyses, validation studies; generally cited in clinical reasoning but not primary guideline anchors. Required: full AI-REPORT plus mandated human verification of every numerical claim.

Tier 4 — Guideline-eligible synthesis

Manuscripts likely to enter clinical guidelines, formularies, or computerized decision support. Required: Tier 3 disclosures plus an independent statistical reproduction and a registered statement of clinical scope.

3.1 Decision rule

We operationalize tier assignment via a 7-question decision tree:

1. Quantitative claim?           No -> T1
2. Patient-level data?           No -> T2
3. Inferential statistic?        No -> T2
4. Recommendation explicit?      No -> T3
5. Guideline-cited domain?       No -> T3
6. Reversibility (high/low)?     High -> T3, Low -> T4
7. Sample size > 5000?           Used only for tie-breaking on T3/T4.

In validation against expert tier assignments ($n = 200$ manuscripts), the tree reaches Cohen's $\kappa = 0.81$ against a panel of 4 clinicians.

4. Empirical Mapping

We scraped 482 AI-authored medical preprints from medRxiv 2023-2025 with explicit AI-authorship disclosure or with metadata indicating $\geq 30$ AI-generated text.

Aggregating T3+T4: 22% of the corpus, with only 6% (combined) reporting prospective clinical validation of the AI-mediated steps.

5. Expected-Error Reduction

We model adoption of tier-conditional review requirements with three error rates: $p_e^{\text{auto}} = 0.052$ for AI-only verification, $p_e^{\text{single-human}} = 0.018$ for one human reviewer, $p_e^{\text{dual}} = 0.006$ for two-reviewer adjudication. Mapping these to the corpus and weighting by the clinical weight $w$ from a separate elicitation, we estimate that universal Tier 3+ enforcement reduces expected patient-relevant errors by 71%.

def expected_error_reduction(corpus_tiers, w, p_auto, p_single, p_dual):
    base = sum(p_auto * w[t] for t in corpus_tiers)
    new = sum(
        (p_dual if t >= 3 else p_single) * w[t] for t in corpus_tiers
    )
    return 1 - new / base

6. Discussion

A frequent counter-argument is that gating Tier-3+ submissions on stricter validation will simply push them to lower-tier venues, eroding overall quality. We are sympathetic but observe that AI-authored Tier-3 manuscripts already cluster on a small set of preprint servers, where venue-level adoption is feasible.

A second concern is the burden of dual-reviewer adjudication. We estimate the marginal cost at roughly </span>$420 per Tier-3 manuscript, a tractable expense compared with the per-manuscript value of avoided downstream errors.

7. Limitations

The expected-error-reduction estimate depends on weights $w$ elicited from a 12-clinician panel; alternative weights shift the headline 71% to a 95% credible interval of $[58$. The 6% prospective-validation figure is an upper bound on what authors disclosed; under-reporting is plausible.

8. Conclusion

AI authorship in medicine is heterogeneous in risk and demands a graduated response. RX-RISK is a practical, reproducible scaffolding for that response, and the empirical mapping suggests it would bind on a meaningful and non-trivial fraction of current submissions.

References

1. Topol, E. (2019). Deep Medicine.
2. Liu, X. et al. (2024). Reporting Guidelines for AI in Healthcare: SPIRIT-AI/CONSORT-AI.
3. Sallam, M. (2023). ChatGPT Utility in Healthcare Education, Research, and Practice.
4. WHO (2024). Ethics and Governance of AI for Health.