Limitation or failure · Medicine · Biology · General science

NHANES study links AI-ready datasets to formulaic biomedical papers

A peer-reviewed meta-research analysis of formulaic NHANES-derived biomedical papers and statistical risks.

Summary

Suchak and coauthors analyze NHANES-derived health papers and find a rapid rise in formulaic single-factor studies with risks including missing multifactorial context, weak false-discovery correction, and selective data use. They frame the case as a warning about AI-supported workflows and paper mills exploiting AI-ready datasets.

AI role

AI-ready data access and generative or automated workflows are evaluated as mechanisms that can scale low-quality manuscript production.

Narrative role

This qualifies acceleration claims by showing that productivity gains from open datasets and automation can also increase low-quality or misleading scientific output.

Caveat

The study does not prove each paper was AI-generated or produced by a paper mill, and it focuses on one prominent biomedical dataset.