Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 12/23/2004
Publication Date: 8/8/2004
Citation: Martino, K., Marks, B., Campos, D., Tamplin, M.L. 2004. Contribution of primary and secondary model uncertainty to the robustness of a broth-based microbial growth model for listeria monocytogenes in meat and poultryproducts. Meeting Abstract. P100
Technical Abstract: Broth-based microbial growth models in the USDA Pathogen Modeling Program (PMP) predict log counts with 95% confidence intervals. These confidence intervals reflect the uncertainty in secondary models, but not the inherent uncertainty in the estimates of the primary model parameters. Quantitative microbial risk assessment requires knowledge of the total uncertainty of the model. In this study, the total uncertainty of a broth-based growth model for Listeria monocytogenes was quantified, with specific application to meat and poultry products. Nonlinear regression was performed with the original broth-based data, and the parameters for the Gompertz equation were estimated. Response-surface secondary models were generated as a function of pH, temperature, salt and nitrite. Model robustness was then quantified via a robustness index (RI), which is defined as the ratio of the standard error of prediction (SEP) to the standard error of calibration (SEC). The SEP and SEC are the root mean square errors (RMSE) of the model against the validation (meat-based) and calibration (broth-based) data, respectively. The maximum, mean, and minimum RMSE of the primary model fit for the broth-based L. monocytogenes data were 7.78, 0.36, and 0.01 log (CFU/mL), respectively. The R2 of the secondary regressions for the B and M Gompertz parameters were 0.44 and 0.82, respectively. The total uncertainty of the model was 1.61 log (CFU/mL); 22% of this was due to the primary model uncertainty, and 78% was due to the secondary model. The RI ranged from 0.87 to 3.99.