Harvard researchers developed PDGrapher, an AI-driven model that predicts combinations of therapeutic targets capable of reversing disease states at the cellular level. Unlike traditional single-target drug discovery, this causality-inspired approach identifies gene perturbations needed to transform diseased gene expression toward healthy profiles, potentially expediting the design of effective combination therapies for complex conditions. PDGrapher could accelerate drug development by targeting multiple disease drivers simultaneously, moving beyond one-size-fits-all approaches.