DRUG-DRUG INTERACTION PREDICTION USING MACHINE LEARNING

Abstract

<p>The identification of <a href="https://ijetrm.com/issues/files/Jul-2025-22-1753152688-JULY37.pdf" target="_blank" rel="noopener">drug-drug interactions (DDIs) </a>is critical for patient safety and optimizing therapeutic<br>outcomes, yet it remains a significant challenge within clinical and pharmacological domains. Existing<br>methodologies often grapple with limitations such as labour-intensive manual review processes and reactive postmarket surveillance. This paper introduces a novel integrated framework for drug interaction analysis, combining<br>a direct knowledge-based lookup mechanism with a machine learning-driven predictive component. Our system<br>employs a Boost classifier, meticulously trained on a curated dataset of drug pair relationships, to discern potential<br>interactions. Drug entities are represented using Label Encoding for model input, yielding robust classification<br>performance, evidenced by an accuracy and an F1-score. Furthermore, a intuitive graphical user interface (GUI)<br>has been developed to facilitate rapid access to detailed descriptions of known interactions, drawing from a<br>comprehensive database. This integrated solution provides a powerful and streamlined approach to uncovering<br>and interpreting drug interactions, thereby contributing to enhanced patient care and supporting more precise<br>pharmacological interventions.</p>