Recent Advances in QSAR Modeling for Predicting Drug Activity Using Machine Learning

Quantitative Structure-Activity Relationship (QSAR) is a computational strategy used to correlate molecular structures with their biological activities through statistical models. It utilizes molecular descriptors (numerical values representing a compound’s electronic, steric, hydrophobic, and topological properties) to predict activity. When the focus is on physical or chemical properties rather than biological activity, the method is referred to as Quantitative Structure-Property Relationship (QSPR). QSAR/QSPR approaches are essential for prioritizing compounds for synthesis and testing, enabling researchers to predict and optimize the activity of new drug candidates in silico before laboratory validation. Recent advancements in QSAR modeling have significantly enhanced its predictive accuracy, robustness, and applicability domain. In this article, research studies published between ۲۰۱۹ and ۲۰۲۵ employing various machine learning methods are reviewed. The number of molecular descriptors and molecules used in each model is summarized for different drug families.

کلیدواژه ها:

Quantitative structure-activity relationship (QSAR) ، Drug Activity ، Half-maximal inhibitory concentration (IC۵۰) ، Drug Design ، Machine learning