Improving drug repositioning accuracy using non-negative matrix tri-factorization

Drug repositioning is a transformative approach in drug discovery, offering a pathway to repurpose existing drugs for new therapeutic uses. In this study, we introduce the IDDNMTF model designed to predict drug repositioning opportunities with greater precision. The IDDNMTF model integrates multiple datasets, allowing for a more comprehensive analysis of drug-disease associations. We evaluated the IDDNMTF model using various combinations of datasets and found that its performance, as measured by AUC, AUPR, and F1 scores, improved with the inclusion of more data. This trend underscores the importance of data diversity in strengthening predictive capabilities. Comparatively, the IDDNMTF model demonstrated superior performance against the NMF model, solidifying its potential in drug repositioning. In summary, the IDDNMTF model offers a promising tool for identifying new therapeutic uses for existing drugs. Its predictive accuracy and interpretability are poised to accelerate the transition from bench to bedside, contributing to personalized medicine and the development of targeted treatments.