International Science Community Association Research Journal of Physical Sciences 14 2 2026 08 4 Correlation and Regression Analysis of VL(G) and Its Transformed Forms for Predicting Physicochemical Properties of Breast Cancer Drugs 14 24 EN Yombouno Faya Maurice 1 Département de Physique. Université de N’Zérékoré, BP 50, N’Zérékoré. Guinée Bayo Ibrahima 2 Centre de Recherche Marine et Côtière, Conakry, Guinée Summy PéLououéa 3 Département de Gestion des Ressources Naturelles. Université de N’Zérékoré. BP 50, N’Zérékoré, Guinée Keita Oumar 4 Département d’Hydrologie. Université de N’Zérékoré. BP 50, N’Zérékoré, Guinée Agbazo Noukpo 5 Département d’Hydrologie. Université de N’Zérékoré. BP 50, N’Zérékoré, Guinée Agrawal Richa 1 Department of Physics, G. N. Khalsa College, Matunga, Mumbai, India Urs Bharathi Shettahalli Nagaraje 1 Department of Mathematics, Government Science College, Chitradurga, Karnataka, India Krishnamurthy Srikumar 2 Department of Mathematics, School of Engineering, Dayananda Sagar University, Bangalore, Karnataka, India > Research Paper 2026 5 20 2026 08 4 As one of the commonly diagnosed cancers worldwide, breast cancer necessitates efficient computational approaches for drug characterization and property prediction. In this study, the predictive ability of the topological descriptor VL(G) and its transformed forms, namely √(VL(G)), [VL(G)]^2, 1/VL(G), and 1/√(VL(G)), was investigated QSPR analysis of selected breast cancer drugs. The molecular structures were represented as molecular graphs, and the selected physicochemical properties were examined through correlation and simple linear regression analysis. The correlation study revealed that VL(G), √(VL(G)), and [VL(G)]² exhibited strong positive relationships with the considered physicochemical properties, whereas the reciprocal forms 1/VL(G) and 1/√(VL(G)) showed inverse correlations with comparatively weaker predictive strength. Consequently, detailed regression analysis was performed only for VL(G), √(VL(G)), and [VL(G)]², producing statistically significant models with high coefficients of determination for most properties. Comparative assessment indicated that VL(G) and [VL(G)]² demonstrated comparatively stronger predictive efficiency for several physicochemical parameters, while √(VL(G)) also exhibited satisfactory performance. The findings establish that VL(G) and its transformed forms effectively capture structural information of breast cancer drugs and can serve as reliable molecular descriptors in QSPR investigations, thereby contributing to graph-theoretical modelling and computational prediction of physicochemical properties in medicinal chemistry. Copyright@ International Science Community Association