Answers are grounded in this paper and cite their sources.
Protocol addendum registered before any of the analyses it describes are run. It covers the revision of a methodological systematic review (meta-research study) of a probability sample, on how data-driven prediction models in food analysis, food safety and nutrition are validated and reported (resubmission of Foods manuscript foods-4598454). The original probability sample of PubMed records was screened and coded with an AI instrument before this registration; its protocol and deviations are reported retrospectively. This addendum fixes in advance the analyses added after peer review: re-coding of the 542 included studies under a corrected grading rule, instrument bridges, a follow-up of the records whose full text was not retrieved, a food-safety search arm, a Scopus sample, a recall audit of excluded records, the estimators, weights and variance methods, and the sensitivity analyses. Randomisation seeds come from the first NIST Randomness Beacon pulse at least 10 minutes after this record's publication time. The SHA-256 of every frozen instrument, search string and script is listed in Appendix A of the addendum; the ZIP holds the 45 registered files with SHA256SUMS.txt.
No takes yet. Share an insight, caveat, or question.
Batayhi et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: