Abstract
The development of effective anticancer drugs remains a central objective in pharmaceutical research. In recent years, topological indices (TIs) have gained considerable attention for their ability to numerically represent molecular structures and support predictive modeling in cheminformatics. This study aims to explore the potential of recently introduced Sombor topological indices and their entropy-based extensions within the framework of quantitative structure-property relationship (QSPR) modeling. The study will focus specifically on anticancer compounds, utilizing graph theory and edge partition approach. A comprehensive Python-based computational framework was developed to compute the relevant topological descriptors and entropy measures. The calculated indices were then integrated with statistical regression and machine learning techniques to construct and evaluate QSPR models to predict characteristics such as boiling point, molar refractivity, heavy atom count, exact mass, flash point, and polarizability. A curated dataset of anticancer agents was employed to ensure data reliability and chemical diversity. Comparative regression analyses indicate that Sombor indices exhibit stronger predictive performance and higher statistical significance than their entropy-based counterparts. These findings highlight the promise of Sombor indices as reliable molecular descriptors for QSPR modeling and powerful tools in the cheminformatics-guided drug discovery process.