In this study, we performed an external validation of Imagene AI’s AI-based NSCLC multi-gene biomarker panel using a blinded test set of more than 2,000 samples, including samples from two clinical trials. We showed that the H&E NSCLC prediction panel achieved consistent performance across multiple key genomic alterations, with AUCs for some of the genes reaching >0.8, and for some, even as high as 0.92–0.93. As we continue to expand the panel, we also demonstrated the feasibility of extending this approach beyond genomic alterations to ADC-related IHC targets, including HER2 expression. These results demonstrate the feasibility of predicting genomic alterations directly from routine H&E images and highlight the potential of this approach to help address some of the challenges patients face on the path to the treatment most suitable for them, as well as supporting clinical trial enrollment.