Modeling and Validation of the Effects of Amino Group Concentrations in Food on the Growth of Escherichia coli

Predictive models for bacterial growth developed on the basis of experimental data obtained from culture media often yield different results from observations in actual foods. Although this discrepancy may be due to differences in compositional characteristics, food structure, and other factors, the impacts on bacterial behavior have not yet been quantified and modeled mathematically. This study first aimed to quantify the effects of amino group concentrations on the growth kinetics of Escherichia coli. A predictive model incorporating the effect of the amino group concentration was subsequently developed, and its potential for improving prediction accuracy in foods was verified. The growth kinetics of E. coli ATCC 25922 were examined at 37 °C in a protein mixture comprising albumin (0.001-30% (w/w)) and phosphate-buffered saline. The maximum specific growth rate (μ_max) and maximum population density (N_max) estimated by the Baranyi and Roberts models were successfully described as equations of the amino group concentration in the form of Monod's model (Monod, 1949)and logarithm, respectively. The developed μ_max equation was further incorporated into the square-root type μ_max model developed by Ross (2003) to improve the predictive robustness. The model performance was validated using the experimentally obtained changes in E. coli numbers over time in actual foods. The root mean squared error (RMSE) of the model incorporating amino group concentration was better (RMSE = 0.652) than that of the model without amino group concentration (RMSE = 0.681). Notably, for lettuce, the prediction accuracy was significantly improved with the model incorporating amino group concentration (RMSE = 0.661) compared to the model without it (RMSE = 1.015). The developed model incorporating the effect of the amino group concentration indicated the potential to reduce the discrepancy between observed bacterial growth in actual foods and model predictions depending on the food type.