Researchers unveiled KGLGANSynergy, a knowledge‑graph and local/global attention network that increases accuracy in predicting drug synergy across compound pairs. The lead sentence: the model integrates chemical, genomic, and pathway knowledge into a graph attention framework and outperformed prior methods on benchmark synergy datasets. The paper details how combining local and global attention enhances interpretability and prediction of combination effects, offering a computational tool to prioritize drug pairs for preclinical testing. This advance may streamline combination discovery and reduce experimental screening burden in early drug development.