Machine learning-assisted optimization of green-synthesized CuO nanoparticles incorporated into Girardinia diversifolia fibers for antibiofilm applications

Optimizing the green synthesis of nanoparticles (NPs) to achieve high yield and controlled size is crucial for developing functionalized biomaterials with antibiofilm activity against multidrug-resistant bacteria. In this study, a Box-Behnken design was adopted to optimize the normalized yield (NY) and particle size (PS) of CuO nanoparticles synthesized using Girardinia diversifolia leaf extract (GLE), considering salt concentration, pH, and GLE. Hyperparameter-optimized artificial neural network (ANN) models, trained on Gaussian noise-augmented datasets, achieved excellent predictive performance for both NY (R 2 : 0.978, R 2 CV : 0.950) and PS (R 2 : 0.999, R 2 CV : 0.998). Genetic algorithm-based multi-objective optimization suggested an optimal salt concentration of 127.65 mM, pH of 9.99, and GLE dose of 41.81% v/v to balance NY (1.00) and PS (27.04 nm). Although the observed NY (0.98 ± 0.06) in the validation experiment under these optimized conditions closely matches the predicted value, the observed PS (63.3 ± 0.6 nm) still differs noticeably. The bioactive metabolites in GLE known to complex and stabilize Cu 2+ ions were identified by GC-MS analysis. CuO NPs synthesized under optimized conditions, when incorporated into G. diversifolia fibers, offered improved NP loading and antibiofilm activity.