Submitted to: International Journal of Food Science and Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/1/2018
Publication Date: 6/17/2018
Citation: Oscar, T.P. 2018. Neural network model for growth of Salmonella Typhimurium in brain heart infusion broth. International Journal of Food Science and Technology. https://doi.org/10.1111/ijfs.13856.
Interpretive Summary: Salmonella are a leading cause of foodborne illness in the United States and throughout the world. A common scenario that leads to foodborne illness is one in which a pathogen cross-contaminates a ready-to-eat (RTE) food during meal preparation followed by temperature abuse that allows the pathogen to grow to a high dose. Computer models that predict the growth of pathogens as a function of variables (temperature, pH) from the current and previous environment are valuable tools for assessing the safety of RTE food that is cross-contaminated and then temperature abused. In the current study, a model was developed that predicts the growth of Salmonella in laboratory broth as a function of previous pH, time, temperature, and pH. Once this model is validated for RTE food, it will provide the food industry with a valuable new tool for assessing food safety in real-time without the need for expensive microbiological tests.
Technical Abstract: Models that predict growth of Salmonella as a function of variables in the current and previous environment are valuable tools for assessing the safety of food. Therefore, the current study was undertaken to develop a model for growth of Salmonella Typhimurium in brain heart infusion broth as a function of previous pH (5.7 to 8.6), temperature (15 to 40°C), pH (5.2 to 7.4), and time. Viable count data (log CFU/ml) were collected and modeled using a neural network approach. The variable impacts were 2.4% for previous pH, 29.0% for temperature, 4.9% for pH, and 63.7% for time. The proportion of residuals in an acceptable prediction zone (pAPZ) from -1 (fail-safe) to 0.5 log CFU/ml (fail-dangerous) were 0.965 (1,061/1,100) for dependent data and 0.939 (386/411) for independent data for interpolation. A pAPZ = 0.7 indicated that the model provided predictions with acceptable accuracy and bias and thus, the model was successfully validated.