Harvard Medical School researchers have developed PDGrapher, an artificial intelligence model that improves drug discovery by predicting combinatorial therapeutic targets capable of reversing disease states in cells. This approach identifies genes that revert diseased cells to healthy function and predicts optimal single or combined treatments, aiming to accelerate development and overcome limitations of traditional one-target drug design. The model is freely available, signaling a potential paradigm shift toward AI-driven drug design addressing complex multi-pathway diseases.