Eccentricity-based topological indices of QSPR modelling using anti-cancer drugs

In chemical graph theory, topological indices are essential for relating physicochemical properties and molecule structures. This study predicts the properties of 23 anti-cancer drugs by combining 13 eccentricity-based topological indices obtained from hydrogen-depleted molecular graphs with regression approaches. Among these, eccentricity-based topological indices are particularly effective at capturing molecular graphs' global structural features. Predictive models are created utilizing linear, ridge, and elastic net regression techniques after the obtained indices are related with specific physicochemical properties. The accuracy of the model was assessed using R ², RMSE, and leave-one-out cross-validation (LOOCV). The EGA index exhibits the most robust predictive accuracy of the examined indices. This work shows that eccentricity-based topological indices provide reliable and informative indications for QSPR modelling and the effective characterization of molecular structures.