Research Information System
17th The International Symposium on Health Informatics and Bioinformatics, İstanbul, Turkey, 18 - 20 December 2024, pp.1
The high cost and lengthy timelines in traditional drug discovery underscore the critical need for innovative and efficient approaches. Extensive clinical trials, rigorous regulatory approvals, and substantial financial investments are the main challenges in bringing new drugs into the market. Conversely, computational methods offer a promising avenue for mitigating these challenges and accelerating drug discovery. Drug repositioning (repurposing existing drugs for new therapeutic indications) represents a particularly effective strategy. This approach leverages the already-established safety profiles of approved drugs, significantly reducing development time, costs, and risks associated with traditional drug discovery. This research utilizes a novel computational method for drug repositioning, employing a translational entity embedding-based neural network model. The model is trained on the comprehensive biomedical knowledge graph provided by the Semantic Medline Database, learning to effectively represent the complex relationships between biomedical entities such as drugs, diseases, genes, and proteins. The model's performance is rigorously validated using repoDB, a widely recognized gold-standard dataset for evaluating drug repositioning methods. Technically, the model learns to minimize the vector distance between semantically related entities. This distance-based approach allows for the prediction of potential drug-disease associations. Shorter distances between entity embeddings indicate a higher likelihood of a therapeutic relationship. The presented study offers a computationally efficient and powerful tool for exploring drug candidates, which expedites drug discovery and leads to new treatments for various diseases. For instance, the model yielded a potential treatment relation between Rolipram and Edema. The results show the model's potential to significantly accelerate the drug development process and improve the efficiency of drug repositioning efforts.