Machine Learning-Based Prediction of Metal Oxide Nanoparticle Cytotoxicity Using Physicochemical and Biological Characteristics
DOI:
https://doi.org/10.69980/cnr4f940Keywords:
Metal Oxide Nanoparticles, Cytotoxicity, Machine Learning, Nanotoxicology, Explainable Artificial Intelligence, Physicochemical PropertiesAbstract
Metal oxide nanoparticles are widely employed in biomedical, pharmaceutical and nanotechnological fields, but their toxicity is highly dependent on the physicochemical characteristics, exposure conditions and biological context. Machine-learning methods offer a practical pathway towards incorporating these diverse features into predictive toxicity assessment. The data set was curated and analysed using a subset of metal oxide nanoparticles after endpoint filtering, duplicate removal, leakage control, categorical encoding, scaling, and stratified train-test partition. The following models were tested in this study using cross validation and independent test performance: Logistic Regression, Random Forest, Support Vector Machine, Gradient Boosting and XGBoost. Logistic Regression had the best test ROC-AUC value of 0.8165 and the most consistent cross-validation discrimination, while Gradient Boosting had the highest test accuracy of 0.7627, a test recall of 0.9762, and a test F1-score of 0.8542. Using SHAP analysis, oxidative potential, dissolution rate, impurity content, release of Zn²⁺ ions, zeta potential, particle size, nanoparticle identity and exposure-related characteristics were identified as key factors contributing to the prediction of cytotoxicity in the ExAI analysis. The biological relevance of the computational results was further corroborated by the strong correlation between the biological processes of membrane damage, apoptosis, necrosis and reduction in cell viability. The modelling framework excluded direct toxicity-response indicators from the primary predictor set, which minimized the leakage of targets and gave a more realistic estimate of predictive performance. The findings highlight the importance of interpretable machine learning in nanotoxicity screening, material prioritisation and safer-by-design development of nanoparticles, and the need for larger harmonised datasets and external independent validation.
References
1. Ahmadi, M., Ayyoubzadeh, S. M., & Ghorbani-Bidkorpeh, F. (2024). Toxicity prediction of nanoparticles using machine learning approaches. Toxicology, 501, 153697.
2. Banaye Yazdipour, A., Masoorian, H., Ahmadi, M., Mohammadzadeh, N., & Ayyoubzadeh, S. M. (2023). Predicting the toxicity of nanoparticles using artificial intelligence tools: a systematic review. Nanotoxicology, 17(1), 62-77.
3. Canta, M., & Cauda, V. (2020). The investigation of the parameters affecting the ZnO nanoparticle cytotoxicity behaviour: a tutorial review. Biomaterials science, 8(22), 6157-6174.
4. Cao, J., Pan, Y., Jiang, Y., Qi, R., Yuan, B., Jia, Z., ... & Wang, Q. (2020). Computer-aided nanotoxicology: risk assessment of metal oxide nanoparticles via nano-QSAR. Green chemistry, 22(11), 3512-3521.
5. Ding, R., Li, Y., Yu, Y., Sun, Z., & Duan, J. (2023). Prospects and hazards of silica nanoparticles: Biological impacts and implicated mechanisms. Biotechnology advances, 69, 108277.
6. Gakis, G. P., Aviziotis, I. G., & Charitidis, C. A. (2023). Metal and metal oxide nanoparticle toxicity: moving towards a more holistic structure–activity approach. Environmental Science: Nano, 10(3), 761-780.
7. Janjua, T. I., Cao, Y., Kleitz, F., Linden, M., Yu, C., & Popat, A. (2023). Silica nanoparticles: a review of their safety and current strategies to overcome biological barriers. Advanced Drug Delivery Reviews, 203, 115115.
8. Ji, Z., Guo, W., Wood, E. L., Liu, J., Sakkiah, S., Xu, X., ... & Hong, H. (2022). Machine learning models for predicting cytotoxicity of nanomaterials. Chemical Research in Toxicology, 35(2), 125-139.
9. Li, J., Wang, C., Yue, L., Chen, F., Cao, X., & Wang, Z. (2022). Nano-QSAR modeling for predicting the cytotoxicity of metallic and metal oxide nanoparticles: A review. Ecotoxicology and Environmental Safety, 243, 113955.
10. Manuja, A., Kumar, B., Kumar, R., Chhabra, D., Ghosh, M., Manuja, M., ... & Prasad, M. (2021). Metal/metal oxide nanoparticles: Toxicity concerns associated with their physical state and remediation for biomedical applications. Toxicology Reports, 8, 1970-1978.
11. Masarkar, A., Maparu, A. K., Nukavarapu, Y. S., & Rai, B. (2024). Predicting cytotoxicity of nanoparticles: a meta-analysis using machine learning. ACS Applied Nano Materials, 7(17), 19991-20002.
12. Na, M., Nam, S. H., Moon, K., & Kim, J. (2023). Development of a nano-QSAR model for predicting the toxicity of nano-metal oxide mixtures to Aliivibrio fischeri. Environmental Science: Nano, 10(1), 325-337.
13. Naz, S., Gul, A., Zia, M., & Javed, R. (2023). Synthesis, biomedical applications, and toxicity of CuO nanoparticles. Applied microbiology and biotechnology, 107(4), 1039-1061.
14. Nikzamir, M., Akbarzadeh, A., & Panahi, P. Y. (2021). An overview on nanoparticles used in biomedicine and their cytotoxicity. Journal of Drug Delivery Science and Technology, 61, 102316.
15. Peng, T., Wei, C., Yu, F., Xu, J., Zhou, Q., Shi, T., & Hu, X. (2020). Predicting nanotoxicity by an integrated machine learning and metabolomics approach. Environmental pollution, 267, 115434.
16. Roy, J., & Roy, K. (2022). Risk assessment and data gap filling of toxicity of metal oxide nanoparticles (MeO x NPs) used in nanomedicines: a mechanistic QSAR approach. Environmental Science: Nano, 9(9), 3456-3470.
17. Savarkar, Soham; Gibson, Jason; Vasanthakumar, Balasubramanian; Moudgil, Brij M.; Hennig, Richard (2025). Machine Learning-Ready Dataset for Cytotoxicity Prediction of Metal Oxide Nanoparticles. Hosokawa Powder Technology Foundation. Dataset. https://doi.org/10.50931/data.kona.29672717.v1
18. Sayed, G. I., Alshater, H., & Hassanien, A. E. (2024). Predicting the potential toxicity of the metal oxide nanoparticles using machine learning algorithms: GI Sayed et al. Soft Computing, 28(17), 10235-10261.
19. Yu, H., Zhao, Z., & Cheng, F. (2021). Predicting and investigating cytotoxicity of nanoparticles by translucent machine learning. Chemosphere, 276, 130164.
20. Zhang, F., Wang, Z., Peijnenburg, W. J., & Vijver, M. G. (2023). Machine learning-driven QSAR models for predicting the mixture toxicity of nanoparticles. Environment international, 177, 108025.
21. Zhou, Y., Wang, Y., Peijnenburg, W., Vijver, M. G., Balraadjsing, S., Dong, Z., ... & Fan, W. (2024). Application of machine learning in nanotoxicology: a critical review and perspective. Environmental Science & Technology, 58(34), 14973-14993.
