Supplementary Material for: Claim-indexed auditing of design–conclusion consistency in biomedical AI studies

The framework proposes a claim-indexed audit procedure to grade the relation between what biomedical AI studies claim and what their designs can support. This supplementary document includes: Appendix A: Minimum Audit Table Template and coding guidelines. Appendix B: Extended worked examples (Cases A–D) with verbatim extractions and re-scoped wording. Appendix C: Detailed search strategy and parameters. Appendix D: Decision rules for boundary cases (e.g., Unverifiable vs. Unsupported; handling of DCA and interpretability claims). Appendix E: Functional comparison of existing appraisal instruments (e.g., TRIPOD+AI, PROBAST+AI). Appendix F: Regulatory, procurement, and post-market applications. Note: No new primary data were generated or analyzed in this methodological study. All quantitative figures cited in the main text are attributed to previously published studies.