Development of a Secondary Model for the Growth of Salmonella Enterica in Food Applying Artificial Neural Network and Combase (Database)

A secondary model of Salmonella growth in food was developed in the form of an artificial neural network (ANN). The dependent variables were microbial growth parameters (lag phase duration and maximum growth rate), and the independent variables included temperature and pH, along with food chemical components (Na+, carbohydrate, sugar, protein, lipid, and water contents). Three hundred and eight data were collected from ComBase and FoodData Central, an official open database, for the microbial growth parameters and food components. Data cleansing was performed to exclude missing values and outliers and select only sterilized or non-sterilized food data. The ANN was constructed based on a back-propagation-based learning algorithm using a commercial program. To prove the accuracy of the model, Salmonella enterica ser. Enteritidis (ATCC 13076) was inoculated into high-protein, high-fat, high-carbohydrate foods, such as pork, fresh cream, and sweet potato, and into low-protein, low-fat, low-carbohydrate foods, such as radish sprout, paprika, and carrots. As a result of comparing the growth parameters by experiment and prediction, the ANN model showed higher prediction accuracy than the model considering only temperature and pH.