Title: General regression neural network model for behavior of Salmonella on chicken meat during cold storage Author
Submitted to: Journal of Food Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: February 26, 2014
Publication Date: May 1, 2014
Citation: Oscar, T.P. 2014. General regression neural network model for behavior of Salmonella on chicken meat during cold storage. Journal of Food Science. 79(5):M978-M987. Interpretive Summary: During cold storage, the number of human bacterial pathogens like Salmonella on chicken meat may remain same, increase, or decrease depending on the type of chicken meat, storage time, and temperature. Accordingly, behavior of Salmonella on different types of chicken meat stored at frozen and refrigerated temperatures was investigated. The number of Salmonella on chicken meat was found to stay the same during frozen storage at temperatures from 18 to 32F and during refrigerated storage at temperatures from 32 to 50 F for up to 8 days. However, the number of Salmonella on chicken meat stored at refrigeration temperatures from 50 to 61F was found to increase with the highest increase on thigh meat, followed by skin, and then breast meat. The data were used to develop a neural network model that predicts the number of Salmonella on chicken breast, skin, and thigh meat stored at temperatures from 18 to 61 F for 8 days. The model will be entered into the USDA, Pathogen Modeling Program where it will be available online for use by regulatory agencies, the food industry, and consumers in the United States and throughout the world.
Technical Abstract: The objective of this study was to investigate and model the behavior of Salmonella on different types of chicken meat during frozen and refrigerated storage. Portions (0.69 to 0.83 g) of chicken meat (breast, skin, or thigh) were inoculated with a single strain (ATCC 700408) of Salmonella Typhimurium definitive phage type 104 (DT104; 2.8 log/portion) followed by storage for 0 to 8 days at -8, -4, 0, 4, 8, 10, 12, 14, or 16 C. A general regression neural network (GRNN) model was developed using commercially available software programs (Excel and NeuralTools). Performance of the GRNN model was considered acceptable when the proportion of residuals (observed – predicted) in an acceptable prediction zone (pAPZ) from -1 log (fail-safe) to 0.5 log (fail-dangerous) was equal to or greater than 0.7. Growth of S. Typhimurium DT104 on chicken meat portions was only observed at 12, 14, and 16 C and differed (P less than 0.05) among types of chicken meat. Growth was highest on thigh, intermediate on skin, and lowest on breast. At lower temperatures (-8 to 10 C), S. Typhimurium DT104 remained at initial levels throughout 8 days of storage. The GRNN model had acceptable performance for all combinations of independent variables except for skin stored for 8 days at 12 C and for skin stored for 6 or 8 days at 14 C. Results of this study indicated that it is important to include type of chicken meat as an independent variable in the model and that the model can be used with confidence to assess and manage effects of cold storage deviations on the risk of illness from chicken contaminated with S. Typhimurium DT104 with the exception of extended storage times on skin held at 12 or 14 C.