A Neural Network on Biomedical Knowledge Graph for the Prediction of Drug-Drug Interactions

Abstract

Effective drug combination prediction is crucial for the achievement of drug discovery, but it is a challenging task due to drug drug interactions and potential adverse drug reactions. This study presents an innovative technique named DDI-KGAT, which employs attention mechanisms to identify crucial characteristics and interrelationships between drugs and various entities, including targets and genes, using a knowledge graph-based strategy. By leveraging associated relations in the knowledge graph, our model adeptly captures drugs and their potential surroundings, thus extracting semantic relations and higher order structures of graph. The KEGG dataset is utilized in evaluating the model's effectiveness, and is compared to other state of the art techniques. The outcomes demonstrate that KGAT outperforms these methods. Additionally, our approach has several advantages, including simplicity, interpretability, and low-dimensional complexity, making it a favorable tool for accelerating the drug discovery and development. By identifying novel drug combinations with improved efficacy and safety profiles, our approach has the capability to improve the patient outcomes and support safer drug development. Our study highlights the potential of attention mechanisms in knowledge graph-based drug combination prediction, and we believe that KGAT framework has the potential to be a valuable foundation for future research in this field.