Skip to main content
ARS Home » Northeast Area » Wyndmoor, Pennsylvania » Eastern Regional Research Center » Residue Chemistry and Predictive Microbiology Research » Research » Research Project #440376

Research Project: Advanced Methods for Predictive Modeling of Bacterial Growth and Survival in Foods

Location: Residue Chemistry and Predictive Microbiology Research

2022 Annual Report

Objective 1: Utilize one-step dynamic modeling and Bayesian analysis for prediction of growth and survival of foodborne pathogens throughout the supply chain. Objective 2: Utilize logistic modeling for determination of growth and no-growth boundary of high-risk pathogens in ready-to-eat foods. Objective 3: Utilize finite element analysis for prediction of bacterial growth and survival during food processing. Objective 4: User-friendly tools for predictive modeling.

A new dynamic approach will be developed and optimized to simulate and predict the growth and survival of major foodborne pathogens in meat and poultry products exposed to complex changes in the environmental conditions during heating, cooling, and storage. The research will utilize an advanced computational framework and probabilistic Monte Carlo simulation to analyze the dynamic changes in the population of foodborne pathogens, and will develop an expert decision support system to assist the food industry and regulatory agencies in making scientifically sound food safety decisions for products of concern. This project will continue to improve and upgrade the USDA Integrated Pathogen Modeling Program (IPMP) for data analysis, and develop a new data analysis tool, IPMP Global Fit, that minimizes the global residual errors for curve-fitting of growth and survival curves.

Progress Report
For Objective 1, research was conducted to evaluate the effect of four different factors that may potentially affect the growth of C. perfringens in cooked cured beef. These factors include sodium nitrite (curing agent, 100 – 200 ppm), phosphate (water holding agent, 0 – 0.5%), sodium chloride (flavoring agent, 1 – 3%), and erythorbate (curing accelerator, 0 – 547 ppm). Previous studies suggest that high concentrations of phosphate and sodium chloride may inhibit the growth of C. perfringens during cooling and isothermal incubation. However, conflicting reports about the effect of sodium nitrite and erythorbate on the growth of C. perfringens have been found. Therefore, a 3-block Box-Behnken response surface design was developed to evaluate these factors on the growth of C. perfringens using a 15-h cooling profile (USDA Food Safety and Inspection Service-FSIS Appendix B Option 1.3, 54.4 – 26.7 degrees C in 5 h and 26.7 – 4.4 degrees C in 10 h). This design includes a total of 27 runs, specifically designed to evaluate the interactive effect of nitrite and erythorbate along with salt and phosphate. Inoculated samples, with a mixture of ingredients, are kept at 4.4 degrees C (40 degrees F) for 24 h to mimic industrial curing. The samples are heated to a final temperature of 75° C and then cooled in a programmable precision refrigerated incubator. In addition, the study would apply a new method, the maximum likelihood analysis, for data analysis to estimate the kinetic parameters. A R package, LAM (Latent Variable Models), is being evaluated for such an application. The LAM package contains Markov Chain Monte Carlo (MCMC Bayesian analysis) procedures for latent variable modeling with a particular focus on multilevel data. This package has been demonstrated to be capable of estimating the parameters in isothermal growth models using MCMC analysis, but it needs to be adapted to dynamic modeling. This study also intends to use optimal experimental design (OED) for data collection and more effective model development. For this purpose, different OED tools are being evaluated, including one Python package (NLoed) and two R packages (PFIM 4.0 and PopED), which have been developed for nonlinear optimal experimental design in systems biology. These tools allow numerical analysis of differential equations and maximize the Fisher Information Matrix (FIM) for parameter estimation. However, these packages are originally developed for pharmacokinetic studies. The dynamic models used in predictive microbiology are not included in these packages. Therefore, they must be modified with custom-defined functions with R or Python for predictive modeling in microbiology. The research is in progress with growth and survival data being collected to establish the methodology and FIM for predictive microbiology. It is anticipated that this research will develop an innovate approach for multi-factor predictive modeling of microorganisms in foods. More practically, it may provide a more conclusive answer to the effect of curing accelerator on growth of C. perfringens in cooked meats during cooling and lead to more effective strategies to control the growth of this pathogen. For Objective 2, a study was conducted to screen the concentrations of salt, nitrite, and sodium tripolyphosphate (STPP) on the growth and no-growth of C. perfringens in Shahidi Ferguson Perfringens agar. The objective of the screening was to identify the ranges of concentrations of these additives that permit the growth and no-growth of C. perfringens for further examinations in cured meat products. The levels of these additives in commercial ready-to-eat cured cooked meat products (1-2.5% salt, 100-200 ppm nitrite, and 0-0.5% of phosphate) were tested. A central composite design was used to select 15 combinations of various levels of salt, nitrite, and STPP for examining the growth and no-growth of C. perfringens under its optimal growth temperature of 46 degrees C. The growth and no-growth responses in 96-well microtiter plates were visually observed and optically measured. Preliminary results showed that three combinations, 1.75% salt-150 ppm nitrite-0.25% STPP, 2.5% salt-200 ppm nitrite-0.25% STPP, and 2.5% salt-200 ppm nitrite-0.5% STPP, delayed the growth of C. perfringens. These three combinations in ground meat and additional additive levels are being examined. For Objective 3, a numerical method based on finite difference analysis (FDA) with finite volume energy balance scheme is being developed to simulate the change in the temperature of foods during heating and cooling. This method may be the most effective method to simulate the temperature history and is capable of solving very “stiff” nonlinear computational problems (partial differential equations, or PDE) involving a phase change such as freezing/thawing during food processing. A computer program written in Fortran was developed to solve the PDEs. IMSL (International Mathematics and Statistics Library), a specialized FORTRAN numerical analysis library, is applied to solve the complex heat transfer process involving temperature-dependent physical properties. A new instrument for measuring the thermal conductivity and diffusivity of food has been set up to evaluate the effect of temperature on these parameters. A data acquisition has been set up to continuously monitor and record the temperature history during food processing. This research is expected to develop a method and computational tool to accurately predict the temperature of food during processing, transportation, and storage, which is essential for evaluating the growth and survival of foodborne pathogens in foods exposed to a dynamic change in temperature through the supply chain and managing the outbreaks of foodborne illness. Specifically, this program may be used by USDA Food Safety and Inspection Service to predict the growth and survival of L. monocytogenes and C. perfringens in large-mass cooked ready-to-eat meats. For Objective 4, a study is being conducted to examine the growth of Bacillus cereus in enzyme-modified liquid eggs. This study is a subtopic in a USDA FSIS-ARS Interagency Agreement. The objective of this study is to determine if B. cereus would grow in liquid eggs during various enzyme treatment processes, which may occur at elevated temperatures such as 50 degrees C for hours, potentially allowing this microorganism to grow and produce toxins. The current objective is to examine the growth of B. cereus in liquid egg yolk (LEY) during treatment with phospholipase A2 (PLA2). Since 50 degrees C is close to the upper temperature limit for most microorganisms, B. cytotoxicus, which is the most heat-tolerant strain among different phylogenetic groups of B. cereus, was selected for this study. B. cytotoxicus, producing Cytotoxin-K, was originally isolated in Europe and was involved in fatal foodborne outbreaks. B. cytotoxicus NVH 391-98 is a type strain and is currently available in DSMZ, the German Collection of Microorganisms. We have received this strain after getting approval from USDA APHIS. This strain has been propagated in the lab and is shown to grow in Mannitol Egg Yolk Polymyxin agar (MYP agar), with colonies morphologically typical of B. cereus on this agar. Two preliminary experiments have been performed to evaluate the growth of B. cytotoxicus in LEY during PLA2 treatment at 50 degrees C. Preliminary results indicate that B. cytotoxicus could grow prolifically at 50 degrees C, indicating a potential risk of growth of this microorganism or other heat-adapted mesophilic groups of B. cereus during enzyme treatment. Dynamic experiments, which expose the samples to a wide temperature range (10 – 55 degrees C), are being conducted to understand the growth kinetics of this microorganism in LEY/PLA2 samples. Preliminary dynamic experiments also suggest that this microorganism may grow well under suitable temperature ranges (20 – 52 degrees C). The current plan is to use different dynamic temperature profiles to test the growth limits and observe the effect of temperature. Purified PLA2 is currently used for enzyme treatment at 10 unit/g of LEY. However, an industrial enzyme will be used in the future. Once the data are collected from different dynamic temperature profiles, one-step dynamic analysis will be used to develop a growth model. The model will be validated using LEY and then further tested in liquid whole egg under isothermal and dynamic conditions.

1. Fermenting and drying can control the growth of Listeria Monocytogenes in meat sausages. Listeria monocytogenes is a significant foodborne health hazard in many products and may survive and grow when making fermented meat sausages. ARS scientists at Wyndmoor, Pennsylvania, developed a process for simultaneous fermentation and drying (SFD) of meat sausages using a starter culture of lactic acid bacteria (LAB) under 30°C and 76% relative humidity. This process allows LAB to grow normally in meat sausages, while effectively inhibiting L. monocytogenes. Mathematical models were developed to describe the competition between LAB and L. monocytogenes, showing the suppression of L. monocytogenes by LAB. Mathematical models were also developed for describing the changes in pH and water activity in meat sausages during SFD. These models can be used to help the food industry safely produce semi-dry and dry fermented meat sausages, while preventing the growth of L. monocytogenes.

Review Publications
Huang, L., Hwang, C., Liu, Y., Renye Jr, J.A., Jia, Z. 2022. Growth competition between lactic acid bacteria and Listeria monocytogenes during meat fermentation – A Mathematical Modeling. Food Control. 158(2022):111553.