General regression neural network model for behavior of Salmonella on chicken meat during cold storage

A study was undertaken to investigate and model behavior of Salmonella on chicken meat during cold storage at constant temperatures. Chicken meat (white, dark, or skin) portions (0.75 cm(3)) were inoculated with a single strain of Salmonella Typhimurium DT104 (2.8 log) followed by storage for 0 to 8 d at -8, 0, 8, 12, 14, or 16 °C for model development and at -4, 4, 10, or 14 °C for model validation. A general regression neural network model was developed with commercial software. Performance of the 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 logs (fail-dangerous) was ≥ 0.7. Growth of Salmonella Typhimurium DT104 on chicken meat was observed at 12, 14, and 16 °C and was highest on dark meat, intermediate on skin, and lowest on white meat. At lower temperatures (-8 to 10 °C) Salmonella Typhimurium DT104 remained at initial levels throughout 8 d of storage except at 4 °C where there was a small (0.4 log) but significant decline. The model had acceptable performance (pAPZ = 0.929) for dependent data (n = 482) and acceptable performance (pAPZ = 0.923) for independent data (n = 235). Results indicated that it is important to include type of meat as an independent variable in the model and that the model provided valid predictions of the behavior of Salmonella Typhimurium DT104 on chicken skin, white, and dark meat during storage for 0 to 8 d at constant temperatures from -8 to 16 °C.

A model for predicting behavior of Salmonella on chicken meat during cold storage was developed and validated. The model will help the chicken industry to better predict and manage this risk to public health.