{
  "abstract": "Cross-validation is routinely used to compare performance in biomedical artificial intelligence. Standard tests ignore correlation across cross-validation folds, inflating false-positive rates. In a PRISMA-guided meta-analysis of 184 studies (impact factor [≥] 15) across 30 biomedical fields, 97% use invalid tests. Among studies with abstract-level claims supported by invalid tests, 59% rely on a spurious comparison: one that loses significance after correcting for this correlation. On average, regaining significance requires a 43% larger effect. Extending these findings to the broader literature, we estimate that spurious comparisons occur in nearly two in three studies and support abstract-level claims in nearly one in three studies. Simulations confirm false-positive rates paradoxically approach 100% when cross-validation is repeated to improve stability. We introduce SHARP, a redesign of cross-validation, which best balances false-positive control and power among 13 benchmarked tests. These results reveal widespread fragility in biomedical artificial intelligence and offer a practical route to valid comparisons.",
  "abstract_shown": true,
  "affiliations": [
    "National University of Singapore"
  ],
  "article_type": "New results",
  "arxiv": null,
  "author_total": 44,
  "authors": [
    {
      "name": "T. Zeng"
    },
    {
      "name": "H. Li"
    },
    {
      "name": "S. Zhang"
    },
    {
      "name": "Y. Q. Tan"
    },
    {
      "name": "F. Tian"
    },
    {
      "name": "C. Orban"
    },
    {
      "name": "L. An"
    },
    {
      "name": "W. Che"
    },
    {
      "name": "J. Cheng"
    },
    {
      "name": "J. S. X. Chong"
    },
    {
      "name": "N. Dehestani"
    },
    {
      "name": "Z. Dong"
    },
    {
      "name": "X. Li"
    },
    {
      "name": "Z. Li"
    },
    {
      "name": "M. J. R. Lim"
    },
    {
      "name": "Y. Lin"
    },
    {
      "name": "Q. Ling"
    },
    {
      "name": "Z. Ling"
    },
    {
      "name": "X. Z. Low"
    },
    {
      "name": "S. Mansour L."
    },
    {
      "name": "K. K. Ng"
    },
    {
      "name": "T. T. Nguyen"
    },
    {
      "name": "L. Q. R. Ooi"
    },
    {
      "name": "S. Pande"
    },
    {
      "name": "X. Qian"
    },
    {
      "name": "J. Ruan"
    },
    {
      "name": "Z. Wang"
    },
    {
      "name": "Y. Xie"
    },
    {
      "name": "C. Zhang"
    },
    {
      "name": "Y. Zhang"
    },
    {
      "name": "K. Patil"
    },
    {
      "name": "L. Parkes"
    },
    {
      "name": "E. Dhamala"
    },
    {
      "name": "S. Chopra"
    },
    {
      "name": "A. Zalesky"
    },
    {
      "name": "A. J. Holmes"
    },
    {
      "name": "S. Eickhoff"
    },
    {
      "name": "J. H. Zhou"
    },
    {
      "name": "O. Renaud"
    },
    {
      "name": "N. Dosenbach"
    },
    {
      "name": "K. P. Kording"
    },
    {
      "name": "D. Bzdok"
    },
    {
      "name": "T. Nichols"
    },
    {
      "affs": [
        0
      ],
      "name": "B. T. T. Yeo"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/1r6d2wtzetz6t.bib",
  "cited_by": 1,
  "doi": "10.64898/2026.05.17.724301",
  "fulltext_url": "https://www.biorxiv.org/content/10.64898/2026.05.17.724301v3.full",
  "id": "1r6d2wtzetz6t",
  "issns": [],
  "landing_url": "https://www.biorxiv.org/content/10.64898/2026.05.17.724301v3",
  "language": null,
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "biorxiv",
      "bioinformatics"
    ],
    [
      "openalex",
      "2613|Statistical Methods in Epidemiology"
    ]
  ],
  "openalex": "W7161802591",
  "pdf_url": "https://www.biorxiv.org/content/10.64898/2026.05.17.724301v3.full.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-05-17",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/1r6d2wtzetz6t.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:21:10Z",
      "native_id": "10.64898/2026.05.17.724301",
      "source": "biorxiv",
      "url": "https://www.biorxiv.org/content/10.64898/2026.05.17.724301v3"
    }
  ],
  "subjects": [
    {
      "field": "biology",
      "origin": "biorxiv:bioinformatics",
      "subfield": "bioinformatics"
    }
  ],
  "title": "Spurious model comparisons are widespread in biomedical artificial intelligence",
  "title_text": "Spurious model comparisons are widespread in biomedical artificial intelligence",
  "type": "preprint",
  "updated_at": "2026-10-30T12:21:10Z",
  "url": "https://research.draconitic.com/paper/1r6d2wtzetz6t",
  "venue": "bioRxiv",
  "versions": []
}