A recent multicentre study in Lancet Oncology demonstrates that a multimodal deep learning model can predict Oncotype DX recurrence scores directly from H&E histopathology with high accuracy (AUC ~0.90) and consistent external validation across diverse cohorts. • AUC 0.898 for RS ≥26 • r=0.73 vs genomic score • External AUC 0.86–0.90 • HR 2.6–2.9 for recurrence endpoints Overall, this work represents a significant step toward integrating digital pathology and AI into therapeutic decision-making in breast cancer. Prospective trials are required before clinical implementation. @dviramar2 @TheLancetOncol @OncoAlert 👉 thelancet.com/journals/lanon…