Computational Intelligent Modeling for Evaluating Metal Oxide Nanomaterial Cytotoxicity Against Escherichia coli

Abstract

Background: Metal oxide nanoparticles (MeOxNPs) possess unique physicochemical properties that differentiate them from bulk-sized forms, making them highly versatile for scientific, medical, chemical, and industrial applications. However, their small size and large surface area contribute to considerable chemical reactivity and underlying toxicity, necessitating comprehensive cytotoxicity evaluations. Previous studies have empirically assessed the toxicity of 17 metal oxides, while mathematical modeling has been proposed to establish a relationship between nanoparticle properties and toxicity. Methods: In this study, we introduced a computational modeling approach to predict the cytotoxicity of metal oxide nanomaterials against Escherichia coli as a model organism. A dataset comprising 17 MeOxNPs and their respective cytotoxicity values was collected from the literature and used to train the models. Molecular descriptors, including cation charge, electronegativity, molecular weight, and other factors, were employed to encode cytotoxicity. The models were evaluated using 10-fold cross-validation. Feature selection using the Relief algorithm was performed to identify the most important features for predicting toxicity. Linear regression was then applied as the predictive model. Results: The performance of the models was assessed based on their accuracy in predicting the cytotoxicity of metal oxide nanomaterials. The proposed models demonstrated high prediction capability. Then, we ranked the top 20 features in descending order of importance. Conclusion: The results indicated that the developed models provide a reliable mathematical framework for predicting the cytotoxicity of metal oxide nanomaterials to E. coli. This provides valuable insights for researchers in the field and supports the design of safer nanomaterials.