<ul><li>Drug repositioning aims to find new uses for existing drugs to save time and money in drug development.</li><li>Most current deep learning methods for drug repositioning use graph-based representations, but struggle in new drug scenarios.</li><li>BiBLDR is a new bidirectional behavior learning approach for drug repositioning that captures drug-disease interaction patterns.</li><li>BiBLDR outperforms existing methods in cold-start scenarios and achieves state-of-the-art performance on benchmark datasets.</li></ul>

BiBLDR: Bidirectional Behavior Learning for Drug Repositioning

Discover more