Ultrasound-assisted quercetin treatment for mono- and dual-species biofilm eradication in milk: Modeling and optimization using GA–ANN approach

The presence of mono- and dual-species biofilms in food industry poses a critical threat with respect to food security and safety at a global scale. This study explored a novel synergistic eradication strategy using ultrasonication and quercetin, for the eradication of mono and dual-species biofilms of Escherichia coli and Salmonella Typhimurium in milk. The eradication of biofilms was simulated via Response Surface Methodology (RSM) and a Genetic Algorithm-Artificial Neural Network (GA-ANN), taking treatment parameters like ultrasonic amplitude, time and bioactive concentration as inputs variable. Quercetin alone exhibited a significant dose- and time-dependent biofilm inactivation, with the maximum reduction of 2.49 ± 0.08 log for S. Typhimurium and 2.14 ± 0.14 log for E. coli mono-species biofilms, at 4 mg/mL after 4 h of exposure. However, this efficacy decreased in dual-species biofilms, confirming their enhanced structural integrity and resilience. The GA-ANN models showed a better predictive accuracy (R² > 0.98; IoA > 0.99) compared to RSM, successfully accounting synergistic interactions among all input variables. The optimized parameter obtained from GA-ANN model demonstrated the highest biofilm inactivation (7.19 log reduction) at 70% US amplitude, for 20 min of US-time, and bioactive concentration of 4 mg/ml. These results demonstrated the GA-ANN model's resilience as a predictive and optimization tool for complex biological systems like multispecies biofilms in milk. This synergistic combination of ultrasonication and quercetin as a novel hurdle approach can be utilized for enhancing microbial safety on food processing surfaces as well as in food matrices.