ChemBERTaDDI: Transforming Drug-Drug Interaction Prediction with Transformers and Clinical Insights

The problem of polypharmacy arises when two or more drugs taken in combination cause adverse side effects, even when the use of the drugs individually causes no harm. Drug-drug interactions (DDIs) are a major cause of these reactions, contributing to increased morbidity and mortality. As the potential for harmful DDI grows exponentially, the prediction of drug-drug interactions is increasingly critical for patient safety and effective healthcare management. In this paper, we develop the ChemBERTaDDI framework, which effectively combines clinical domain data, represented by the mono side-effect features with the enriched chemical molecular representations derived from ChemBERTa-77M-MLM , a transformer-based language model. Experiments performed on a benchmark data set show superior performance compared with five state-of-the-art methods: Decagon, DeepWalk, DEDICOM, NNPS, and RESCAL. The evaluation demonstrates that ChemBERTaDDI achieves an F1 score of 0.94 and an AUROC of 0.97 outperforming the baseline architectures and generalizing to new introduced drug compounds.