Answers are grounded in this paper and cite their sources.
Protocol and first amendment of a systematic review and meta-analysis of artificial intelligence models that infer genomic, transcriptomic and protein alterations in breast cancer from routine histopathology, immunohistochemistry and clinical data ("virtual molecular biopsy"), registered with PROSPERO (CRD420261567463). This deposit time-stamps the protocol before study selection. Amendment 1 (dated 10 November 2026, before any record was screened) moves screening to Rayyan with blinding enabled, removes all automated assistance from study selection, adds the sex and the reported race or ethnicity of each validation cohort as extraction items, and corrects the spelling of one author's name. Eligibility criteria, outcomes and the planned synthesis are unchanged. Files: registered protocol; amendment 1; search strategies for all seven databases as run; supplementary appendix including the dated table of protocol changes; prespecified R analysis code. The review data, screening decision logs, extraction dataset, risk-of-bias judgements and the logs of AI assistance will be deposited as a new version of this record at submission.
No takes yet. Share an insight, caveat, or question.
Pavan et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: