Data-Driven Machine Learning to Predict Antibacterial Activity of Cerium-Doped Nanoparticles

Nowadays, nanomaterials are a real alternative for controlling antibiotic-resistant bacteria. Several nanoparticles have shown good performance in inducing bacterial death due to a photoassisted process. This study employs statistical analyses and machine learning (ML) models to investigate the effect of doping ZnO nanoparticles with cerium ions on their antibacterial activity in a dark environment. Incorporating cerium ions into the ZnO matrix was systematically analyzed in terms of structural, morphological, and optical parameters. The incorporation of cerium ions did not modify the crystal structure. These results were correlated with their qualitative and quantitative antibacterial activity against Escherichia coli, Staphylococcus aureus, and Pseudomonas aeruginosa to establish the impact of doping. No significant differences in their antibacterial activity were observed. A maximum of 95% bacterial growth inhibition was observed. ML tools were utilized to model bacterial survival under different conditions. The support vector machine (SVM) model yielded the highest prediction error. In contrast, the extremely random tree model produced an error of only 1.8%, making it an excellent computational tool for decision-making. Using this framework, we conducted attribute importance analysis based on the model, identifying a small subset of parameters as being crucial for generating a precise model. The findings provide light on the most critical characteristics of cerium-doped ZnO nanoparticles’ antibacterial activity.