ISSN print edition: 0366-6352 ISSN electronic edition: 1336-9075 Registr. No.: MK SR 9/7 Published monthly

Topological index-based entropy measures and AI-driven prediction of electrochemical and optical properties of a phenyl-linked diketopyrrolopyrrole-thiophene dimer

Wakeel Ahmed, Anas Raza, Nimra Javed, Shahid Zaman, and Muhammad Danish

Department of Mathematics, University of Sialkot, Sialkot, Pakistan

E-mail: wakeelahmed784@gmail.com

Received: 27 December 2025  Accepted: 17 March 2026

Abstract:

This study presents a comprehensive analysis of the influence of molecular topology on the physicochemical behavior of a selected chemical reaction system by employing a range of degree-based and entropy-based topological indices to quantitatively characterize the molecular structure. The computed topological descriptors were used as predictive variables to model key electrochemical, optical, and physicochemical properties through machine learning algorithms. Regression analysis enabled the identification of the most influential topological indices with clear mathematical interpretability, while the artificial neural network (ANN) model effectively captured nonlinear and higher-order relationships among the descriptors, resulting in improved predictive performance. A comparative assessment demonstrated that ANN provided higher accuracy, whereas regression offered greater transparency and simplicity. The combined application of classical statistical modeling and machine learning highlights their complementary roles and confirms that integrating topological indices with ANN-based modeling enhances the accuracy and reliability of chemical property prediction, offering valuable insights for molecular design and theoretical chemical analysis.

Keywords: Entropy measure; Diketopyrrolopyrrole; Electrochemical properties; Artificial neural network; Curve fitting

Full paper is available at www.springerlink.com.

DOI: 10.1007/s11696-026-04824-3

Chemical Papers 80 (7) 7391–7411 (2026)