ENHANCING DRUG SIDE EFFECT PREDICTION WITH EXPLAINABLE AI FOR MEDICAL HEALTH APPLICATIONS

Abstract

Adverse drug reactions represent a critical challenge in clinical practice, often arising when potential side effects are overlooked during early prescription stages. This study aims to enhance drug side effect prediction by integrating machine learning with Explainable AI (XAI), creating an intelligent and interpretable system for healthcare applications. Using a comprehensive dataset of drug attributes, side effect profiles, and relevant clinical features, the study first evaluates baseline classifiers including Ridge Classifier, Linear SVM, Logistic Regression, and Multinomial Naïve Bayes assessing their performance through accuracy, precision, recall, and F1-score. Extensive Exploratory Data Analysis (EDA) is performed to identify patterns, correlations, and class imbalances, guiding effective feature selection and preprocessing. To improve predictive capability and capture complex, nonlinear relationships in the data, a Multi-Layer Perceptron (MLP) Classifier is proposed. The MLP model demonstrates superior performance compared to traditional methods, effectively recognizing intricate feature interactions and enabling high-accuracy predictions. By combining deep learning with explainability, the system provides transparent, trustworthy insights for clinical decision support systems, promoting safer drug administration and improved patient outcomes. Overall, this work delivers a reliable, interpretable, and high-performing solution for drug side effect prediction, bridging the gap between advanced AI techniques and practical healthcare applications.