Predictive analysis of physicochemical properties using novel weighted Sombor indices and comparative machine learning approach
Discover Chemistry · 29.06.2026 · 3 (1)
DOI: 10.1007/s44371-026-00840-0 ↗Kurz gefasst
Fachartikel in der Zeitschrift Discover Chemistry, online erschienen am 29.06.2026. Verfasst von Zunaira Kosar und 3 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 23 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
The pharmaceutical industry currently faces a “combinatorial explosion” of potential drug-like molecules. Predicting their physical properties quickly is essential for modern drug discovery. Testing millions of theoretical compounds in a physical laboratory exceeds the capacity of traditional laboratory testing. Therefore, researchers need fast mathematical models to screen these chemicals before they are synthesized. Traditional Quantitative Structure-Property Relationship (QSPR) models rely on basic 2D graph theory to transform chemical structures into numerical data. However, standard unweighted topological indices treat all atoms identically, lacking to capture the unique physical differences between atoms like carbon and oxygen. To overcome this, our study introduces a “Physico-Topological” framework using Weighted Sombor Indices, modifying the geometric graph with atomic mass, radius, electronegativity, and ionization energy. We tested 28 novel descriptors on 200 pharmaceutical compounds, comparing classical Multiple Linear Regression (MLR) against advanced machine learning algorithms (Random Forest, XGBoost, ANN). Results show that linear models accurately predict size-dependent properties like Molar Refractivity (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}^{2}\approx\:0.98$$\end{document}). Conversely, electronic properties like Polarizability require the non-linear architecture of Artificial Neural Networks (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}^{2}=\:0.93$$\end{document}). While models achieved moderate accuracy for Boiling Point (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}^{2}=\:0.75$$\end{document}), they failed for Density (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}^{2}
Zitieren
APA
Kosar, Z., Khan, M. S., Zaman, S., & Ahmed, W. (2026). Predictive analysis of physicochemical properties using novel weighted Sombor indices and comparative machine learning approach. Discover Chemistry, 3(1). https://doi.org/10.1007/s44371-026-00840-0
Vancouver
Kosar Z, Khan MS, Zaman S, Ahmed W. Predictive analysis of physicochemical properties using novel weighted Sombor indices and comparative machine learning approach. Discover Chemistry. 2026;3(1). doi:10.1007/s44371-026-00840-0
BibTeX
@article{Kosar2026,
author = {Kosar, Zunaira and Khan, Muhammad Shehroze and Zaman, Shahid and Ahmed, Wakeel},
title = {Predictive analysis of physicochemical properties using novel weighted Sombor indices and comparative machine learning approach},
journal = {Discover Chemistry},
year = {2026},
volume = {3},
number = {1},
doi = {10.1007/s44371-026-00840-0}
}
EUVIMED