INTELLIGENT DRUG SAFETY MONITORING WITH INTERPRETABLE MACHINE LEARNING MODELS

Abstract

Adverse drug reactions pose significant risks in clinical settings, particularly when side effects are overlooked during the early stages of prescription. To mitigate these risks, this study focuses on enhancing drug side effect prediction using machine learning techniques integrated with Explainable AI (XAI) for medical applications. The primary objective is to develop an intelligent, interpretable system that not only predicts potential side effects but also provides transparency in the decisionmaking process, thereby fostering trust among healthcare professionals. A comprehensive dataset comprising drug attributes, side effect profiles, and associated clinical features was used for model training and evaluation. Initial experimentation involved various baseline classifiers, including Ridge Classifier, Linear Support Vector Machine (SVM), Logistic Regression, and Multinomial Naïve Bayes. These models served as benchmarks and were evaluated based on accuracy, precision, recall, and F1-score. Extensive Exploratory Data Analysis (EDA) was performed to uncover patterns, correlations, and class imbalances within the dataset, aiding in informed feature selection and preprocessing. To improve prediction accuracy and enable complex pattern recognition, a Multi-Layer Perceptron (MLP) Classifier was proposed as the advanced model. As a deep learning algorithm, the MLP demonstrated superior performance in capturing nonlinear relationships among features— capabilities that traditional models often lack. This makes the solution suitable for real-world deployment in Clinical Decision Support Systems (CDSS), ensuring safer drug administration and better patient outcomes. The project contributes to the field of medical AI by delivering a highperforming, interpretable, and reliable solution for drug side effect prediction, effectively bridging the gap between complex AI models and practical healthcare applications.