Background: Predicting genomic alterations from routine hematoxylin and eosin (H at the patch-level, prediction of APCSNV reached an AUROC of 0. 916, and prediction of KRASSNV reached an AUROC of 0. 811 on the held-out test set. Conclusions: In a heterogeneous clinical gene-panel setting, pathology foundation models can provide strong baseline genomic-prediction signals without additional fine-tuning. We propose a practical, deployment-oriented two-stage workflow: rapid slide-embedding screening to prioritize robust representations and candidate endpoints, followed by patch-level training for high-value tasks where additional performance gains and interpretable regions are clinically worthwhile.
Ma et al. (Wed,) studied this question.