Hoang, Danh-TaiDinstag, GalShulman, Eldad D.Hermida, Leandro C.Ben-Zvi, Doreen S.Elis, EfratCaley, KatherineSammut, Stephen-JohnSinha, SanjuSinha, NeelamDampier, Christopher H.Stossel, ChaniPatil, TejasRajan, ArunLassoued, WiemStrauss, JuliusBailey, ShaniaAllen, ClintRedman, JasonBeker, TuvikJiang, PengGolan, TaliaWilkinson, ScottSowalsky, Adam G.Pine, Sharon R.Caldas, CarlosGulley, James L.Aldape, KennethAharonov, RanitStone, Eric A.Ruppin, Eytan2025-12-172025-12-17WOS:001261446000001PubMed:38961276ORCID:/0000-0003-2640-4196/work/193722137ORCID:/0000-0002-2725-4209/work/193722465https://hdl.handle.net/1885/733795797Advances in artificial intelligence have paved the way for leveraging hematoxylin and eosin-stained tumor slides for precision oncology. We present ENLIGHT-DeepPT, an indirect two-step approach consisting of (1) DeepPT, a deep-learning framework that predicts genome-wide tumor mRNA expression from slides, and (2) ENLIGHT, which predicts response to targeted and immune therapies from the inferred expression values. We show that DeepPT successfully predicts transcriptomics in all 16 The Cancer Genome Atlas cohorts tested and generalizes well to two independent datasets. ENLIGHT-DeepPT successfully predicts true responders in five independent patient cohorts involving four different treatments spanning six cancer types, with an overall odds ratio of 2.28 and a 39.5% increased response rate among predicted responders versus the baseline rate. Notably, its prediction accuracy, obtained without any training on the treatment data, is comparable to that achieved by directly predicting the response from the images, which requires specific training on the treatment evaluation cohorts.Hoang et al. developed a deep-learning framework called ENLIGHT-DeepPT that predicts therapy response based on imputed transcriptomics and shows predictive power across patient cohorts and cancer types.This work was partially supported by grant DP190103402 from the Australian Research Council (D.-T.H., E.A.S.) and by the Intramural Research Program of the National Institutes of Health (NIH), National Cancer Institute (NCI), Center for Cancer Research (CCR) (E.D.S., S.S., N.S., E.R.). This work used the supercomputational resources of the Australian National Computational Infrastructure (AUNCI) and the Australian National University Merit Allocation Scheme (ANUMAS). D.-T.H. would like to thank T. Bui for the helpful discussion.25en© 2024 The Author(s)Artificial-intelligenceProstate-cancerBiopsies2024-07-0310.1038/s43018-024-00793-285197370611