THE USE OF MACHINE LEARNING IN PREDICTING DRUG INTERACTIONS

Abstract

Drug-drug interactions are crucial in drug research. Nevertheless, they may also elicit unfavorable responses in patients, leading to severe repercussions. The manual identification of drug-drug interactions is a laborious and costly process, necessitating the immediate use of computer-based approaches to address this issue. Computers can find medication interactions using two methods: by recognizing established drug interactions and by forecasting undiscovered drug interactions. This study provides an overview of the advancements in machine learning for predicting unfamiliar medication interactions. Out of these techniques, the literature-based method stands out because it integrates the DDI extraction method with the DDI prediction method. Initially, we provide the commonly used databases. Subsequently, we provide a concise description of each approach and conclude by summarizing the merits and drawbacks of several prediction models. Lastly, we will examine the difficulties and potential of machine learning techniques in forecasting medication interactions. This study intends to provide valuable assistance to researchers interested in advancing bioinformatics algorithms for predicting drug-drug interactions (DDI).