Interpretable machine learning models for QSAR-based prediction of anti-Salmonella typhi activity

This study aimed to develop a robust machine learning (ML)-based quantitative structure-activity relationship (QSAR) model to identify potential drug candidates active against multidrug-resistant Salmonella typhi. A curated ChEMBL-derived dataset was assessed for modelability, yielding a high MODI value of 0.89. A hybrid feature selection workflow was applied to retain 20 chemically interpretable molecular descriptors, and eight diverse ML classifiers were systematically trained and benchmarked. The Support Vector Machine (SVM) model achieved the highest performance (MCC = 0.61, ROC-AUC = 0.90) on the hold-out test set. Overall, rigorous ML-QSAR modeling offers a reliable and efficient framework for virtual screening and prioritization of novel anti-S. typhi agents discovery.