International Conference on Predictive Modelling in Food Paris 2013 September 16-20 www.icpmf8.org Software Fair Booklet PARIS, SEPTEMBER 16-20, 2013
CONTENTS THE SOFTWARE FAIR STORY... 1 THE SOFTWARE FAIR RUNNING... 3 SOFTWARE FAIR DESCRIPTION... 5 BASELINE SOFTWARE TOOL... 6 FDA RISK... 8 FILTREX... 10 COMBASE... 12 MICROHIBRO... 14 MICROBIAL RESPONSES VIEWER... 16 NIZO PREMIA... 18 PMM LAB... 20 FISHMAP... 22 GINAFIT... 24 PREDICTION OF MICROBIAL SAFETY IN MEAT PRODUCTS... 26 SYM PREVIUS... 28 GROPIN... 30 FOOD SPOILAGE AND SAFETY PREDICTOR... 32 DAIRY PRODUCT SAFETY PREDICTOR... 34 LISTERIA MEAT MODEL... 36
The Software Fair story To encourage the use of software and the transfer from researchers to food business operators, the organization committee of ICPMF8 planned to organize a special session dedicated to the demonstration of software for predictive microbiology. Two sessions were planned: 1. A plenary oral presentation session to present the top 5 best applications, 2. A demonstration session in a speed dating friendly format where all the applicants will have a chance to present their tool to the Software Fair delegates A call for contribution was launched on January 2013. It was published on the conference website www.icpmf8.org and sent to a list of 30 software developers. Applicants were asked to fill in an online questionnaire to give a short description of their software, its main applications, its target audience The software fair call was successful as 16 applications were received, exceeding the target number fixed by the organizing committee. A selection panel of 10 national experts working in the field of predictive microbiology was set. Each expert proposed a ranking for the 16 software according to the questionnaire results and to key criteria such as scientific relevance, international visibility, ergonomics and innovation. A dedicated meeting was organized and experts rankings were gathered to select the top 5 software for the oral presentation session. The organizing committee finally decided to give the opportunity to all the software to take part to the demonstration activities in a friendly speed-dating format with 6 sessions of 20 minutes each. Enjoy the Fair! Olivier COUVERT Mariem ELLOUZE Fanny TENENHAUS-AZIZA 1
National selection panel Jean-Chistophe Augustin ENVA Louis Coroller LUBEM / UBO Olivier Couvert LUBEM / UBO Juliette Dibie Barthélemy AgroParisTech / INRA Mohammed El Jabri ADRIA Développement Mariem Ellouze IFIP Laurent Guillier ANSES Lydie Soler INRA Valérie Stahl AERIAL Fanny Tenenhaus-Aziza CNIEL Software Fair presenters Presenter(s) Company / Institution Baseline Software Tool A. Valero University of Cordoba FDA-iRISK Y. Chen U.S. Food and Drug Administration FILTREX J-P. Gauchi INRA ComBase J. Baranyi and D. Marin Institute of Food Research Microhibro F. Perez-Rodriguez University of Cordoba MRV, Microbial Responses Viewer S. Koseki National Food Research Institute NIZO Premia M. Verschueren NIZO food research PMM-Lab M. Filter Federal Institute for Risk Assessment FISHMAP B. Alfaro AZTI-Tecnalia GInaFiT Prediction of microbial safety in meat products A. H. Geeraerd and L. Haberbeck A. Gunvig KU Leuven Danish Meat Research Institute Sym'Previus N. Desriac ADRIA Développement GroPIN Food Spoilage and Safety Predictor (FSSP) Dairy products safety predictor P.N. Skandamis P. Dalgaard H. Souaifi, F. Tenenhaus-Aziza Agricultural University of Athens National Food Institute (DTU Food) ACTALIA / CNIEL Listeria Meat Model J. Van Impe KU Leuven 2
THE Software Fair running Three demonstration rooms are available at the Software Fair center to host the event (see Software Fair Map) Each software will be assigned a table with 7-8 seats and equipped with a flat screen to facilitate the demonstration activities. Presenters are free to organize their demonstrations but you can of course ask them to take the lead and try the software! After each 20 minutes, a bell will ring and participants will be asked to change tables to take part to another software demonstration. To facilitate the rotations, please be advised that you will have to register to the sessions prior to the Software Fair launch (from 8 am). Registration to the desired software can be achieved on Wednesday, September 18th, 2013 on the provided posters at the registration desk. the Software Fair Center Demonstration Room 1 Demonstration Room 2 Demonstration Room 3 Software Fair Center ComBase, MRV, GroPin, BaseLine, Microhibro. Sym Previus, Dairy Products Safety Predictor, GInaFiT, FILTREX, PMM-Lab, Listeria Meat Model. FDA-iRISK, Prediction of microbial Safety in meat products, Food Spoilage & Safety predictor, FISHMAP, NIZO Premia. 3
FDA-iRISK, Prediction of microbia Food Spoilage & Safety FISHMAP, NIZO Premia. PARIS, SEPTEMBER 16-20, 2013 3:40 PM 6:00 PM Software Fair parallel demonstration sessions Demonstration Room 1 ComBase - Daniel Marin MRV, Microbial Responses Viewer - Shige Koseki GroPIN - Panagiotis Skandamis Baseline Software Tool - Antonio Valero Díaz Microhibro - Fernando Perez Rodriguez Demonstration Room 2 Sym'Previus - Noémie Desriac Dairy products safety predictor - Hajer Souaifi, Frédérique Perrin & Fanny Tenenhaus-Aziza GInaFiT - Annemie Geeraerd & Letícia Haberbeck FILTREX - Jean-Pierre Gauchi PMM-Lab - Matthias Filter Listeria Meat Model - Jan Van Impe 5:50 PM 6:30 PM Demonstration Room 3 FDA-iRISK - Yuhuan Chen, Regis Pouillot & Greg Paoli Prediction of microbial safety in meat products Claus Borggaard & Annemarie Gunvig Food Spoilage and Safety Predictor (FSSP) - Paw Dalgaard FISHMAP - Begoña Alfaro NIZO Premia - Maykel Verschueren & Peter de Jong Free discussion with software developers Software Fair Center ComBase, MRV, GroPin, BaseLine, Microhibro. Demonstration Room 1 Demonstration Room 2 Demonstration Room 3 Sym Previus, Dairy Products Safety P GInaFiT, FILTREX, PMM-Lab, Listeria Mea 4
Software Description Disclaimer : All software descriptions were only formatted into the correct font, size and paragraph style and were not language edited. The descriptions were reprinted as submitted by their authors during the Software Fair call. We accept no responsibility for any language, grammar or spelling mistakes. 5
Baseline Software Tool Company / Institution: University of Cordoba, Spain Announcer(s): A. Valero Access on: www.baselineapp.com Created in 2012, free internet access software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module (simulation only) Inactivation module (simulation only) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix (Seafood, Egg products, Dairy products, Meat products, Vegetables) Micro-organisms Pathogens: 5 species Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 6
Short description: This web-based software tool has been developed within the EU Project of the VII Framework Program named BASELINE Selection and improving of fit-for-purpose sampling procedures for specific foods and risks (Grant Agreement No 222738). Throughout this software users can access to different implemented modules related to predictive modelling of pathogens for various food commodities and estimation/ optimization of sampling plans. In the first module, microbial behavior of a selected category of food borne pathogens can be predicted under different environmental conditions. A selection of implemented predictive growth and inactivation models gathered from scientific literature is included: a) Seafood (Listeria monocytogenes and Vibrio spp.). b) Eggs and egg products (Salmonella spp.). c) Milk and dairy products (Listeria monocytogenes and Escherichia coli). d) Meat products (Salmonella spp., Campylobacter spp. and Listeria monocytogenes). e) Plant products (Escherichia coli O157:H7, Salmonella spp. and Listeria monocytogenes). Regarding model`s structure, the main menu allows the selection of predefined primary models: a) Growth: Baranyi, Gompertz and three-phase linear model b) Inactivation: Three-phase linear and Weibull models Customized secondary models (mainly polynomial, Gamma, Ratkowsky type and cardinal) can be easily introduced by users through an equation editor dialog. Besides, new predictive models were added to the software based on experimental data collected in the BASELINE project. The predictive models can be validated by the user in order to compare predictions and observations through the application of goodness of fit indices. This information and modelling results can be exported in CSV, MS Excel or PDF formats. A personal space is provided for each user, allowing them to save and retrieve their own data for further use. By means of restricted local access, administrators will be able to further develop the software tool by including additional predictive models, microorganisms, food matrices or environmental factors. The Sampling Plans module contains a Generic module where the most common distributions for assessing two- and three- class attributes and variables sampling plans are embedded. Users can easily define the parameters n i.e. number of samples to be taken from a given lot; c maximum allowable number of samples that can accomplish a specific criteria, m and M which are the logarithmic values of the microbial concentration that should not be exceeded. The Practical examples module allows the user the setting of food safety criteria (Performance Objectives) in prevalence and/or concentration terms. Specific sampling plans include a selection of sampling schemes related to the food/risk categories included in BASELINE. Finally, in the Tools module users can estimate different parameters associated to a specific sampling process, according to preliminary criteria fixed in the dialog boxes. This information may be used to evaluate the overall effectiveness of applied interventions by risk managers or food operators. Partners: University of Cordoba Optimum Quality 7
FDA irisk Company / Institution: Food and Drug Administration, USA Announcer(s): Y. Chen Access on: http://foodrisk.org/exclusives/fda-irisk-a-comparative-risk-assessment-tool Created in 2012, free internet access software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database (associated with simulation) Growth module Inactivation module Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 8
Short description: The Food and Drug Administration (FDA) created FDA-iRISK, an interactive, web-based risk assessment tool, to provide regulatory and industry decision-makers with a systematic, faster way of assessing risks in the food supply and predicting the effectiveness of control measures. FDA created this new software application to answer a key question: Which foods and contaminants are public-health priorities? Collaborating with many partners, the agency developed FDA-iRISK into a system with built-in mathematical functions and templates that enable users to - simultaneously rank public-health risks from multiple contaminants, in multiple foods; - calculate how contamination and illness from each would change with interventions and changes in food production/processing/handling practices. FDA collaborated with the Joint Institute for Food Safety and Applied Nutrition, Risk Sciences International, and others to make the tool globally available at FoodRisk.org. Partners: Food and Drug Administration Joint Institute for Food Safety and Applied Nutrition Risk Sciences International Canadian Food Inspection Agency 9
Filtrex Company / Institution: INRA, France Announcer(s): J.P. GAUCHI Access on: www.jouy.inra.fr/mia Created in 2013, Free software / tool to be uploaded on computers Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module (fitting only) Inactivation module Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens: 1 species Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 10
Short description: FILTREX: A New Software for Parametric Identification and Optimal Sampling of Experiments for Complex Microbiological Dynamic Systems by Nonlinear Filtering In the present release of this (free) software, three functionalities are proposed: 1: Parametric identification This parametric identification concerns microbiological dynamic systems, based on primary models (growth or thermal inactivation models). This FILTREX identification functionality is based on the implementation of a new nonlinear particle technique using a convolution kernel approach (Rossi and Vila, 2005, 2006). Let us just recall here that for this efficient particle filtering procedure, the only a priori information needed for the parameters is their respective possible variation ranges. The coding of this functionality in FILTREX has been developped from an open source code of the convolution particle filter (Choquet and Rossi, 2005). 2: Model comparison and selection This second functionality computes the so called Bayes Factor, for deciding which of two models better fits a given set of data (see Vila and Saley, 2009, for details). This Bayes Factor is the ratio of the respective marginal likelihood functions of the two competing models. It is not a genuine statistical test but it has been proved to be one of the best indices for comparing two nonlinear models. Its particle estimation in FILTREX does not need the knowledge of the model likelihoods as required by the usual statistical selection procedures (e.g. Akaïke criterion). 3: Optimal sequential designs Three technics are proposed in this release: - Construction of a Sobol-Saltelli method based approach at starting time of the dynamics, - Construction of a D-optimal design at starting time of the dynamics, - A more sophisticated and powerful technics: an on-line approach based on the SIVIP method (Gauchi and Vila, 2012). References Choquet R. and Rossi V. (2005) Partners: J.-P. Gauchi¹, J.-P. Vila², C. Bidot¹, E. Atlijani¹, L. Coroller³, J.-C. Augustin⁴, P. Del Moral⁵ 1- Institut National de la Recherche Agronomique (INRA), département Mathématiques et Informatique Appliquées (UR3141), Jouy-en-Josas, France. (jean-pierre.gauchi@jouy.inra.fr). 2- Institut National de la Recherche Agronomique (INRA), département Mathématiques et Informatique Appliquées, Montpellier, France. (jean-pierre.vila@supagro.inra.fr) 3- Université Européenne de Bretagne, France. Université de Brest, EA3882 Laboratoire Universitaire de Biodiversité et Ecologie Microbienne, UMT 08.3 Physi opt, IFR148 ScInBioS, Quimper, France. (louis. coroller@univ-brest.fr) 4- Université Paris-Est, Ecole Nationale Vétérinaire d Alfort, Unité MASQ, Maisons-Alfort, France. (jcaugustin@ vet-alfort.fr) 5- Institut National de Recherche en Informatique et Automatique (INRIA), Equipe ALEA, Bordeaux, France. (pierre.del-moral@inria.fr) 11
ComBase Company / Institution: Institute of Food Research, England Announcer(s): J. Baranyi and D. Marin Access on: www.combase.cc Created in 2004, free internet access software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database (dissociated from simulation) Growth module (simulation and fitting) Inactivation module (simulation and fitting) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens: 15 species Spoilers: 5 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 12
Short description: ComBase is a web-based tool for Predictive Food Microbiology. Its main components are a database of observed microbial responses to a variety of food-related environments and a collection of relevant predictive models. Using an internet interface, users can narrow down their search results to a dataset relevant to their query. Alternatively, ComBase customers may be interested in (and most frequently they are content with) generating predictions provided by the mathematical models developed from selected records in the database. The more than 50,000 records on microbial growth and survival (mostly viable count curves) were donated by research establishments or collated from publications. Systematic experimental design is in the background of those data, which provide the basis of the predictive models accompanying the database. They target the responses of the major food-borne pathogens to environments quantified by temperature, ph, humectants, etc.). ComBase is a result of collaboration between the Institute of Food Research, UK; the University of Tasmania Food Safety Centre (FSC) in Australia; and the USDA Agricultural Research Service (USDA- ARS) in the United States. It provides an electronic repository for food microbiology observations, to make the data and the generated predictive tools freely available and accessible to a wide community interested in quantitative food microbiology. Recently ComBase underwent a major restructure and new features were introduced to make it easier to use, especially for risk assessment. Recently ComBase underwent a major restructuring and new features were introduced to make it easier to use, especially for risk assessment. These include the error estimation of the predicted specific rates, visualising their probability distribution, and separating uncertainty due to lack of information and due to statistical variability when predicting the lag time. These indicators can play crucial role when assessing microbial risk, quantified by an appropriate cost function and driven by the inevitable error in the generated predictions. Partners: Institute of Food Research UK Food Safety Centre University of Tasmania AU USDA Agricultural Research Service, US. 13
Microhibro Company / Institution: University of Cordoba, Spain Announcer(s): F. Perez-Rodriguez Access on: www.microhibro.com Created in 2011, free internet access software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database (associated with simulation) Growth module (simulation only) Inactivation module (simulation only) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens: 6 species Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 14
Short description: The on-line tool is structured in three different modules: 1. Models 2. Validation 3. Risk model The first module consists of a system for predicting microbial growth/inactivation/survival in different food matrices under conditions defined by users. The underlying models allow obtaining estimates of kinetic parameters under different environmental conditions such as maximum growth rate, latency phase, and maximum population density. In addition, the tool displays the kinetic curve, providing the microorganism level at different time intervals. The second model concerns a validation system. Its purpose is to allow the user to compare the selected predictive models with observed values provided by users. Based on different statistical indices and a graphical representation, the model can be evaluated whether or not it makes reliable predictions for a given food on which the user has performed the validation. The third module is an object-oriented tool for the development of probabilistic risk models. Users can design specific risk models based on the combination of different basic processes, thus describing changes in concentration and prevalence of a particular microorganism within a specific food chain. In each process users can define model variables by using point-estimate values or probability distributions. Also, results can be analyzed by applying scenario analysis and/or sensitivity analysis. The module Models contains a database that includes different predictive models for microbial growth and death of different microorganisms, foods, and environmental conditions. The database includes three types of primary models: Three-phase linear model (Buchanan et al., 1997) Gompertz Model Baranyi and Roberts (1994) Model In respect to the secondary models, the module is based on: Gamma (Zwietering et al., 1996) Ratkowsky or square root (Ratkowsky et al.,1982) Cardinal (Rosso et al.,1995) The risk estimates are based on a combination of secondary and primary models contained in this database. In an advanced version, expert users can introduce predictive models based on a expert and intelligent system, enabling that new and updated model can be incorporated in the application easily. In addition, end users can save models, predictions, and reports to be accessed in any moment. The application can be considered as a «test bench» for microbiological predictions and risk assessment for any end user: risk management, risk assessor, or food safety authority staff. Partners: University of Cordoba Spanish Innovation and Technology Ministry 15
Microbial Responses Viewer Company / Institution: National Food Research Institute, Japan Announcer(s): S. Koseki Access on: http: //mrv.nfri.go.jp Created in 2008, free internet access software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database (dissociated from simulation) Growth module Inactivation module Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens: 11 species Spoilers: 10 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 16
Short description: The MRV (Microbial Responses Viewer) is a new database consisting of microbial growth/no growth data derived from ComBase. Furthermore, the specific growth rate of each microorganism was modelled as a function of temperature, ph, and water activity (aw) using three different models. The retrieved data of the growth/no growth data and specific growth rate model are illustrated in a figure. Seventeen kinds of microorganism with growth/no growth data were extracted from all the data in ComBase comprising 30 kinds of microorganism. These data were successfully modelled as a function of temperature, ph, and water activity. The specific growth rate was illustrated using a two-dimensional contour plot with growth/no growth data. Additionally, current MRV enables the comparison of the growth rate as a function of temperature among kinds of bacteria and kinds of food. Partners: ComBase consortium 17
NIZO Premia Company / Institution: NIZO food research, Netherlands Announcer(s): M. Verschueren Access on: no Internet access Created in 1995, Commercial software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module (simulation only) Inactivation module (simulation only) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix (dairy products) Micro-organisms Pathogens: 10 species Spoilers: 10 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 18
Short description: NIZO Premia is a software platform that can be used to design and optimize food production processes with respect to production costs and product quality. The platform comprises a range of process models representing different process units used in production (e.g. membrane separation, heat treatment, falling film evaporation, spray drying, cheese production etc). These process models can be linked to a range of product models that can predict product quality properties as a function of processing conditions and raw material composition. Models for microbial inactivation, adherence of microorganisms (i.e. biofilm formation) and microbial growth (during processing or shelf life) make up an important part of the set of product models. These models can be combined with other models related to product quality aspects such as protein denaturation (including fouling), inactivation of enzymes, viscosity, color etc. By combining relevant process and product models with the NIZO Premia environment it is possible to design and optimize production processes: i.e. minimize production costs e.g. by increasing capacity or maximizing runtime while ensuring product quality and safety. In addition to the functionalities described above, elements within the NIZO Premia platform have also been combined with external software into a QMRA tool. NIZO Premia has been developed in close collaboration with industrial experts and is equipped with a user-friendly GUI. The models have been validated on industrial scale and are used both in R&D and online in production environments in the international dairy industry. Partners: NIZO food research has developed NIZO Premia in close collaboration with experts from the (international) dairy industry 19
PMM-Lab Company / Institution: Federal Institute for Risk Assessment, Germany Announcer(s): M. Filter Access on: https: //sourceforge.net/projects/pmmlab/ Created in 2012, Free software / tool to be uploaded on computers Design for Food business operators Researchers Government Teachers Students Power inside Applications Database (associated with simulation) Growth module (simulation and fitting) Inactivation module (simulation and fitting) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 20
Short description: Title: A community resource for integrated predictive microbial modeling (PMM-Lab) The prediction of growth and inactivation of microorganisms in food matrices is a field of intensive research with increasing relevance to food safety professionals and public authorities. Today, large quantities of experimental and quality control data are available in comprehensive public or proprietary databases. In parallel the mathematical foundation for the generation of predictive microbial models has been developed and is available. Nevertheless there is a gap in easy-to-use, free and transparent software solutions that enable e.g. food safety professionals to apply state-of-the art mathematical modeling concepts to their proprietary data and additionally provide a solution to the issue of integrated data and model management. The software PMM-Lab has been specifically designed to address this gap. It is outlined right from the beginning as community resource to allow broad application and joint development. PMM-Lab already provides many valuable data analysis and modeling features including modules to create, visualize, analyze, save, import, export and deploy predictive microbial models based on experimental data. Moreover, it allows consistent data management for the analyzed experimental data and generated models due to the integrated database. Technically, PMM-Lab is an extension to the open-source data integration and analysis platform KNIME (www.knime.org) from which it inherits highly beneficial properties like modularity, flexibility, scalability and extensibility. PMM-Lab is designed to be used by laboratory scientists, food safety specialists or even modeling experts who wish to distribute their methods to a broad community. Also users trying to answer fundamental research questions based on experimental data collections or looking for structure in their growing data and model collection will find value in this software. Possible application areas are numerous including prediction of shelf life and food spoilage. PMM-Lab has been developed by the Federal Institute for Risk Assessment, Germany and is freely available at https://sourceforge.net/projects/pmmlab/ under the GNU public license. Partners: Authors: Matthias Filter, Christian Thöns, Jörgen Brandt, Armin A. Weiser, Alexander Falenski, Annemarie Käsbohrer, Bernd Appel Institute: Federal Institute for Risk Assessment, Department of Biological Safety 21
FISHMAP Company / Institution: AZTI-Tecnalia, Spain Announcer(s): B. Alfaro Access on: http: //www.azti.es/en/fish-map-software.html Created in 2011, Free software / tool to be uploaded on computers Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module (simulation and fitting) Inactivation module Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix (fish) Micro-organisms Pathogens Spoilers: 5 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 22
Short description: The software developed FISHMAP allows the prediction and the visualization of growth of spoilage bacteria in fish product under modified atmosphere packed (MAP) with different concentration of carbon dioxide (CO2) at constant and fluctuating temperatures. Moreover, the program includes the prediction of growth of spoilage bacteria under air conditions. The spoilage bacteria included in this program are: - Carnobacterium maltaromaticum - Serratia proteamaculans - Shewanella baltica - Yersinia intermedia - Mixed bacteria: (C. maltaromaticum, S. proteamaculans, S. baltica and Yersinia intermedia) Furthermore, the software allows the graphical comparison of experimental growth microbiological data with the respective microbial growth model at either constant or fluctuating temperature. Optionally, the predictions and plots can be saved as an Excel workbook. The predictive models are based on experimental data of bacterial growth obtained at liquid medium under different ambient conditions: temperature interval between 0-20ºC; under CO2 enriched atmospheres (0-100% CO2 v/v, balance nitrogen) and under air condition (21% O2). The growth kinetic parameters were estimated by fitting the model of Baranyi and Roberts (1994). The dependence of the maximum specific growth rate on temperature, CO2 and O2 was described by a cardinal model similar to that described by Mejlholm, et al. (2010). Models included in FISHMAP have been validated comparing observations on naturally contaminated horse mackerel fillets packed under modified atmospheres as well as inoculated with the bacteria strains used to generate the data for model development. These validation studies showed a good performance of models under constant and fluctuating temperature conditions. Partners: AZTI-Tecnalia, Derio, Spain Institute of Food Research, Norwich, UK 23
GInaFiT Company / Institution: KU Leuven - BIOSYST, Belgium Announcer(s): A. Geeraerd Access on: http://cit.kuleuven.be/biotec/downloads.php Created in 2003, Free software / tool to be uploaded on computers Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module Inactivation module (Fitting only) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 24
Short description: GInaFiT is a freeware add-in for Microsoft Excel, aiming at bridging the gap between people developing predictive modelling approaches and end-users in the food industry or research groups not disposing of advanced non-linear regression analysis tools. The tool is useful for testing ten different types of microbial survival models on user-specific experimental data relating the evolution of the microbial population with time. The ten model types are: - classical log-linear curves, - curves displaying a so-called shoulder before a log-linear decrease is apparent, - curves displaying a so-called tail after a log-linear decrease, - survival curves displaying both shoulder and tailing behavior, - concave curves, - convex curves, - convex/concave curves followed by tailing, - biphasic inactivation kinetics, - biphasic inactivation kinetics preceded by a shoulder, - curves with a double concave/convex shape. The models were originally published as Bigelow and Esty (1920), Cerf (1977), Geeraerd et al. (2000), Mafart et al. (2002), Albert and Mafart (2005), Geeraerd et al. (2005) and Coroller et al. (2006). Next to the obtained parameter values, the following statistical measures are automatically reported: standard errors of the parameter values, the Sum of Squared Errors, the (Root) Mean Sum of Squared Errors, the R2 and the adjusted R2. In addition, t4d, the time needed for a 4 log reduction of the initial microbial population, as originally proposed by Buchanan et al. (1993), is also automatically reported (for data sets covering at least 4 decimal reductions). The tool can be used in two ways. On one hand, for end-users having already a qualitative idea of the general shape of their survival curves, the choice for one of the model types is obvious. On the other hand, if the end-user does not have a clear idea yet, two or more of the different model types available can be tested and compared. The time for a 4 decimal reduction can be useful to summarize the information present in a data set, for example, if a common survivor curve shape can not be selected for a range of different conditions tested. Additionally, the tool has some built-in features testing for mis-use, for example, when trying to identify a model with tailing on data not having a tail or when using a too limited number of data points (observations) in comparison with the number of parameters in the model type chosen (the number of parameters ranges from 2 to 5 for the ten model types available). Further illustration on the use of GInaFiT can be found in Geeraerd et al. (2005). Partners: KU Leuven, Leuven, Belgium References: Albert I. and P. Mafart 2005. A modified Weibull model for bacterial inactivation, International Journal of Food Microbiology, 100, 197-211. Bigelow W.D. and J.R. Esty 1920. The thermal death point in relation to typical thermophylic organisms, Journal of Infectious Diseases, 27, 602-617. Buchanan R.L., Golden M.H. and Whiting R.C. 1993. Differentiation of the effects of ph and lactic or acetic acid concentration on the kinetics of Listeria monocytogenes inactivation, Journal of Food Protection 56, 474-478, 484. Cerf O. 1977. A review. Tailing of survival curves of bacterial spores, Journal of Applied Microbiology, 42, 1-19. Coroller L., Leguerinel I., Mettler E., Savy N. and Mafart P. 2006. General model, based on two mixed Weibull distributions of bacterial resistance, for describing various shapes of inactivation curves, Applied and Environmental Microbiology, 72 (10), 6493-6502. Geeraerd A.H., C.H. Herremans and J.F. Van Impe 2000. Structural model requirements to describe microbial inactivation during a mild heat treatment, International Journal of Food Microbiology, 59, 185-209. Geeraerd A.H., V.P. Valdramidis and J.F. Van Impe 2005. GInaFiT, a freeware tool to assess non-log-linear microbial survivor curves, International Journal of Food Microbiology, 102, 95-105. Mafart P., O. Couvert, S. Gaillard and I. Leguerinel 2002. On calculating sterility in thermal preservation methods: application of the Weibull frequency distribution model, International Journal of Food Microbiology, 72, 107-113. 25
Prediction of microbial safety in meat products Company / Institution: Danish Meat Research Institute, Denmark Announcer(s): A. Gunvig Access on: http://3.test.dezone.dk Created in 2006, free internet access software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module (simulation only) Inactivation module (simulation only) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix (Meat products) Micro-organisms Pathogens: 4 species Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 26
Short description: Prediction of microbial safety in meat products by DMRI models DMRI has developed three mathematical models that predict growth, growth/no growth or reductions of pathogens in different meat products. All models are based on data obtained from meat products. The Listeria monocytogenes and C. botulinum models are available on free web-sites. The ConFerm model will be available at the same web-site during summer 2013. All three models can be used by QA employees and risk assessors to evaluate the consequence of changing recipes, e.g. the safety in low salt and no-nitrite products. The advantage of using the DMRI models is the fact that the users can evaluate the safety of a specific product in relation to its content of the given variables. Growth of L. monocytogenes in ready-to-eat meat products: This model predicts the growth of Listeria monocytogenes in ready-to-eat meat products with up to seven different chemical or physical hurdles. Growth data L. monocytogenes were obtained in modified atmosphere packaged (CO2 and N2 mixtures) meat products with up to six other different chemical or physical variables. A total of 446 growth curves with different concentrations/conditions of the variables were used to train an artificial neural network to generate the maximum specific growth rate (µmax) for a combination of the seven variables. The input variables are: Sodium chloride (1.6-6.0 % in water phase); ph (5.4-6.6); added sodium lactate (1-3%); added sodium acetate (0-0.5%); sodium nitrite (0-150 ppm); CO2 (0-100 %); and temperature (2-12 C). The output of the prediction is a growth curve that shows the log L. monocytogenes/g as a function of time (days) in relation to the selected values. Besides the curve the specific growth rate and doubling time are calculated. Growth/no growth of C. botulinum in ready-to-eat meat products: This model predicts growth/no-growth of psychrotrophic C. botulinum in pasteurised meat products packed in modified atmosphere (30% CO2/70% N2) for combinations of storage temperature, ph, NaCl, added sodium nitrite and sodium lactate. Data for developing and training the artificial neural network (ANN) were generated in meat products. A total of 249 growth experiments were carried out in three different meat products with different combinations of the five variables. The inputs of the model are the five variables: Temperature (4-12 C), ph (5.4-6.4), added sodium nitrite (0-150 ppm), sodium chloride (1.2-2.4%) and sodium lactate (0-3% ). The output of the model is presented in a contour plot of an area spanning the allowed values for temperature and water phase salt (WPS). A black dot in the contour plot represents the result of a given combination of the input variables. The depicted area is divided into 3 regions indicating different probabilities of growth (green area= no growth, yellow area = uncertain, red area= growth). ConFerm tool: The ConFerm tool predicts the reduction of Salmonella, verocytotoxigenic E. coli and L. monocytogenes during production of fermented and matured sausages. The data used for developing the ConFerm tool were obtained from fermented sausages produced in a pilot plant. A total of 73 experiments were carried out using sausages with different levels of NaCl in water phase (3.9 6.8 %), sodium nitrite (0 200 ppm) and ph48h (4.3 5.6). The meat mince was inoculated by approx. 105 cfu/g of a multi-cocktail containing Salmonella (3 strains) VT negative E. coli (3 strains) and L. monocytogenes (5 strains). The sausages were fermented at 24 C for 48 hours using 3 different starter cultures followed by maturation at 16 C until obtaining a 15 % or 35 % weight loss. The inputs of the model are: Added salt to the mince (2.8-3.2 %), added sodium nitrite (0-200 ppm), ph48 h (4.3-5.6), phfinal (4.2-5.4), total weight loss (10-35 %), total process time (9-24 days), water content in the final product (24-53 %). The output from each model is a graph showing the count of each bacterium plotted against the weight loss (default) or WPS, which gives the user knowledge of where in the process the highest reduction is obtained. The graphs are supplemented by the estimated total log reduction for Salmonella/VTEC/L. monocytogenes in the final products including the estimated uncertainty. Partners: National Food Institute Technical University (ConFerm) 27
Sym Previus Company / Institution: ADRIA Développement, France Announcer(s): N. Desriac Access on: www.symprevius.org Created in 2003, Commercial software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database (associated with simulation) Growth module (simulation and fitting) Inactivation module (simulation and fitting) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens: 8 species Spoilers: 9 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 28
Short description: Sym Previus : A network, a software and an operational system Sym Previus network gathers expertises in predictive microbiology from major food companies, French technical centres and public research institutes. Based on the most recent concepts in predictive microbiology, Sym Previus proposes an assistance in food safety management. Sym Previus predictive tool is composed of a database and an advanced simulation software. The database provides several informations on microorganisms behaviour in/on foods as well as natural contaminations encountered in foods. The specifically developed querying system, allows a flexible and structured search for a given microorganism, food matrix or food category selected (Buche et al., 2005). Secondly, a user-friendly software simulates microorganisms growth in food matrix or heat destruction after industrial treatment. Sym Previus is an easy way to access predictive microbiology tools for food companies. At the present time, the software describes the effect of temperature, ph, water activity and organic acid on bacterial growth and thermal destruction in a wide range of food categories, such egg products, dairy product, meat, ready-to-eat food and processed meals. Sym Previus allows simulation for most pathogens and bacteria spoilage species. For each species, data from several strains are implemented to account for strains variability: 13 strains for Listeria monocytogenes, 10 for E. Coli (10 strains), 7 for Salmonella and 9 for Bacillus cereus for instance. The main spoilage species are represented at least by one strain (9 species). Finally, simulations based on risk analysis approach could take into account the initial contamination distributions estimated from industrial monitoring data sets, lag-times distributions of individual cells and growth simulations could predict the probability to exceed critical limits of microorganism concentrations in food. All models applied in Sym Previus are published and already validated in a wide variety of food matrices. Applications of this advanced predictive software are therefore widespread. It is nowadays already used to : - predict bacterial growth in several conditions : T, ph, aw and organic acid. - evaluate the growth probability of microorganisms in foods - evaluate contamination level at shelf-life - optimise food formulation (additive, ph, salt) - optimise new process conditions (heat treatment time) - evaluate the impact of cold chain breaks, and test different storage scenarios - help to identify Critical Control Points in a process - simulation of the probability of growth in any food products. Constant improvements and updates ensure Sym Previus state of the art position in the field of simulation tools. Currently these evolutions mainly concern sensitivity analysis to determine which parameters are the key drivers of the model simulation results. Sym Previus is a tool that is constantly evolving with its database and simulation tools being added to through its involvement in national and European research programmers. It is now a reliable tool that complies with the requirements of European regulations (EC) 2073/2005 and 1441/2007 related to the microbiological criteria applicable to foodstuffs. Sym Previus has been evaluated by the French Agency for Food, Environmental and Occupational Health &Safety (ANSES). The commission guidance document (2008) on Listeria monocytogenes shelf-life studies for ready-to-eat food under EC n. 2073/2005 on microbiological criteria for foodstuffs used the probabilistic approach of Sym Previus and screenshot of the software are taken as an exemple. Sym Previus is referenced in the French Direction générale de l Alimentation memorandum n 2010-8062/ March 9th, 2010 on the shelf-life of the food. Sym Previus offers an English and French version available online at www.symprevius.org Partners: Public sector research labs : ENVA, INRA, INPAG ACTIA Technical centers : ADRIA Développement, ACTALIA, AERIAL, IFIP Industries members of UNIR : Bongrain, Danone, Pernod Ricard, Fromageries Bel Public authorities : Ministère de la Recherche, Ministère de l Agriculture et de l Alimentation de la Pêche et des Affaires Rurales (DGAL) 29
Gropin Company / Institution: Agricultural University of Athens, Greece Announcer(s): P. N. Skandamis Access on: www.aua.gr/psomas/gropin/ Created in 2013, Free software / tool to be uploaded on computers Design for Food business operators Researchers Government Teachers Students Power inside Applications Database (associated with simulation) Growth module (simulation only) Inactivation module (simulation only) Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix Micro-organisms Pathogens: 14 species Spoilers: 13 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 30
Short description: An integrated tertiary model called GroPIN is developed in-house using Visual Basic for Applications. The application may serve as a user-friendly and highly transparent predictive modeling data base for kinetic (growth or inactivation) and probabilistic models. It also offers the flexibility of interactive options in selecting the graphical and numerical simulation of models. An unlimited number of mathematical models can be introduced into the database via equation editor, as compared with other applications, where only a limited number of equations are already embedded into the source code and are not (at least not easily) updatable or expandable. The current version of GroPIN has a total of 154 published models for the behavior of 14 pathogens and 13 spoilage organisms, including spoilage and mycotoxigenic fungi, in various foods of plant (e.g., fresh-cut salads, deli salads, berries, juices, etc.) or animal origin (meat and meat products, dairy products). The impact on microbial behavior of a variety of critical and commonly encountered intrinsic (preservatives, organic acids in total or undissociate/dissociate form, salt, aw, nitrates, etc.) and extrinsic (temperature, CO2, pressure, anaerobic conditions) factors is accounted for by the models registered in GroPIN up to date. The microbial responses modelled (i.e., dependent variables) include the maximum specific growth rate, the death rate, the lag phase duration, maximum population density, time to X-log reduction/growth, D-values and the probability of growth. The user may introduce in-house (unpublished) models, too. A search engine has been established for locating and selecting the model of interest. Then the user may select variables and assign values for each variable though list boxes or by direct typing. The simulation of the selected model can be displayed as Response Surface-Contour Plot, Time to x log Response Surface-Contour Plot, growth or inactivation curve, as well as 2D growth/ no growth (probabilistic) interface with potential illustration of up to 3 interfaces (i.e., three levels of the 3rd variable). The following model categories have been included: - Probabilistic models - Growth models - Inactivation-survival models, and - Gamma (Cardinal) Models with interactions ( xsi term based on psi and phi functions, according to Augustin and Carlier., 2000; Le Marc et al. 2002). All kinetic models, including growth or inactivation, plus gamma models with interactions, can be simulated under both static and dynamic conditions. The 154 registered models include 86 growth models, 31 inactivation models, 27 probability of growth models and 10 gamma models with interaction terms. The user can use the available models as a basis for setting performance-, process- or productcriteria, as well as to evaluate the compliance of a product with microbiological criteria regulation. The spirit of the software stems from similar initiatives, such as SymPrevius and COMBASE modeling toolbox. The major innovative features of this software in relation to the state-of-the art are the user-friendliness, the updatable character by the user, the simplicity and functionality (including interactive options) of outputs and the inclusion of all major predictive modeling classes. Partners: Laboratory of Food Quality Control & Hygiene, Department of Food Science & Technology, Agricultural University of Athens 31
Food Spoilage and Safety Predictor Company / Institution: National Food Institute (DTU Food), Technical University of Denmark Announcer(s): P. Dalgaard Access on: http://sssp.dtuaqua.dk Created in 1999, Free software / tool to be uploaded on computers Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module (simulation only) Inactivation module Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix (Various seafood and meat products) Micro-organisms Pathogens: 3 species Spoilers: 3 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 32
Short description: FSSP is a new and expanded version of the Seafood Spoilage and Safety Predictor (SSSP) software which was released for the first time in January 1999. The FSSP software contains new predictive models and new facilities in addition to all the features already available as part of SSSP v. 3.1 from 2009: - relative rate of spoilage models to predict the effect of constant and fluctuating temperature storage conditions on product shelf-life, - models for growth of specific spoilage micro-organisms to predict shelf-life of fresh fish stored aerobically or modified atmosphere packed (MAP) - models to predict histamine formation in marine fin-fish. SSSP software is used by more than 6000 individuals or institutions from 112 different countries. An important part of this success is due to the extensive and accurate Listeria monocytogenes growth and growth boundary model which has been part of SSSP since 2004. This model is recognized world vide and e.g. included in guidelines from food safety authorites on how to evaluate and manage growth of Listeria monocytogenes in ready-to-eat foods. New predictive model and facilities in the FSSP software includes: - A generic growth and growth boundary model: This is a simplified cardinal parameter type model that allows users to enter and save cardinal parameters values. In this way FSSP can be used to predict growth and growth boundary for any microorganism where the cardinal parameter values are known. This generic model can take into account the effect of up to 12 environmental parameters and predictions can be obtained for constant or for dynamic temperatures and ph storage conditions. This generic model is available as part of FSSP for Windows and in a tablet version for ipad. The tablet version has been developed to facilitate the out-of-office-application of predictive microbiology models e.g. when food inspection officers visit food processors or food-outlets and needs to illustrate effects of storage conditions, product composition or hygiene. - A new growth and growth boundary model for lactic acid bacteria in meat and seafood products: This new and extensively validated model includes the effect of 12 environmental parameters. - An expanded model to predict the simultaneous growth of lactic acid bacteria and Listeria monocytogenes in various seafood and meat products. The expanded model includes the effect of 12 environmental parameters and predictions can be obtained for constant or for dynamic temperatures storage conditions. - FSSP can be used in 18 different languages and translation into Japanese and Turkish are new features not previously included in SSSP. The new FSSP software will be available for free from the homepage of the National Food Institute (DTU Food) at the Technical University of Denmark. FSSP becomes available prior to ICPMF8 and can be accessed through the homepage of the available SSSP software (http://sssp.dtuaqua.dk). Partners: Division of Industrial Food Research, National Food Institute (DTU Food), Department of Applied Mathematics and Computer Science (DTU Compute) 33
Dairy Product Safety Predictor Company / Institution: ACTALIA/CNIEL, France Announcer(s): H. Souaifi Access on: https://aqr.maisondulait.fr/ Created in 2012, commercial software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module Inactivation module Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix (Especially cheeses: soft-cheese, semi-soft cheese, hard cheese) Micro-organisms Pathogens: 4 species Spoilers Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 34
Short description: For several years now, the French dairy sector has been developing microbiological risk assessment and exposure assessment models for the safety of dairy products, especially cheeses. To date, when a food business operator has developed an exposure assessment model specific to its product and one microorganism, he has the possibility to perform simulations with parameters of its choice, online, via a user-friendly interface, provided he has subscribed. The model development is a prior step to the online simulations. It is done together with the French technical center for milk and dairy product Actilait and a food business operator. It can last from some months to one year, depending on available data. The model development includes the collection of data regarding physical and chemical parameters all along the process (temperature, ph, water activity and lactic acid), accounting for variability; the statistical analysis of results from challenge tests, if available, to assess optimal growth rate of the microorganism in the considered matrix; the implementation of predictive microbiological models for growth in dynamic conditions (primary model and secondary cardinal models) and destruction, but also models describing cross-contamination; if the cheese is made from raw milk, a statistical analysis of milk routine microbiological results is performed, to assess the initial contamination of the milk (frequency, concentration); if the cheese is made from pasteurized milk, definition of recontamination scenarios is needed, using for instance routine microbiological control of the product and in the environment during process. Generally, the final results of a model are the distribution of the prevalence and the concentration at different steps of the process, until consumption. These results allow then defining sampling plans, verifying if a microbiological criterion is respected, etc. A risk of illness can also be calculated. When finalized, the tailor-made model is loaded in the interface and made available to the food business operator only, via a personal access. When logged in to his account, the user has to download an Excel sheet in which parameters accounted for in the model can be modified. After having changed parameters (for example a different temperature for the ripening phase), the user uploads the new version of the Excel sheet in the interface and starts simulation. When results are available online, he is informed by email. Partners: ACTALIA CNIEL 35
Listeria Meat Model Company / Institution: KU Leuven, Belgium Announcer(s): J. Van Impe Access on: www.cpmf2.be Created in 2012, commercial software Design for Food business operators Researchers Government Teachers Students Power inside Applications Database Growth module Inactivation module Growth / no growth interface Risk assessment module Sensitivity analysis module Media Food matrix (Meat) Micro-organisms Pathogens: 1 species Spoilers: 1 species Growth factors Temperature ph Water activity Lactic acid Others organic acids CO 2 Interactions between factors Modeling approach Deterministic Probabilistic 36
Short description: As part of the control measures for L. monocytogenes, Food Business Operators (FBO) should conduct studies to identify growth potential of L. monocytogenes in products put on the market. Next to the specifications of physicochemical characteristics and available scientific literature, predictive microbiology can be used. Therefore, it is important that existing predictive models are validated for a large category of products and that predictions are compared with results obtained from extensive challenge tests. Based on the data of intrinsic factors provided by 30 Belgian companies and in consultation with these companies, specific recipes were established to prepare model products on pilot scale in standardized conditions. All durability and challenge tests for validation were performed by a BELTEST accredited laboratory (LFMFP-UGent) according to the EU technical guidance document on L. monocytogenes shelf-life studies for RTE foods (EU CRL, 2008). Based on the results of this study, the Listeria Meat Model software package was developed to predict the growth of L. monocytogenes in meat products. The software offers the possibility to compare different products or temperature profiles. The first important input in the software is the initial concentration of L. monocytogenes. The target value immediately after production is absence in 25 g. This corresponds with -1.4 log CFU/g. This initial concentration influences only the time at which the limit of 100 CFU/g will be exceeded. The product characteristics are specified via the most important factors influencing the growth of L. monocytogenes in meat products: salt, ph, residual nitrite, acetic acid/acetate, diacetate, lactic acid/lactate. These data need to be inserted on product basis by the user and are automatically converted to water phase basis by the software, based on specified dry matter. A temperature profile with an unlimited amount of steps can be specified. The model is, however, only validated for the cold chain (temperatures < 15 C). The software allows for higher temperatures but warns the user that this is beyond the validated range for temperature. The amount of dissolved CO 2 at equilibrium, which is dependent on the initial headspace CO 2 concentration, the storage temperature and the gas/product ratio, is a final important factor for products packed under modified atmosphere. A CO 2 calculator in included in the software for those cases where the user does not know the amount of dissolved CO 2. For some inputs (inoculum, salt concentration, ph, dry matter and temperature), uncertainty ranges can be specified. It should be noted that the combination of a lot of uncertainties on the input values could lead to a very large confidence interval around the predicted growth. The output of the software is, on the one hand, a graphical representation on the growth of L. monocytogenes as a function of time and, on the other hand, a numerical output. This latter gives: - the growth potential, the difference between the cell count at the end of shelf life and the initial Listeria concentration, - the time to reach 2 log CFU/g (legal criterion) - the tolerance on the day of production. This tolerance is determined by the calculated growth potential and gives to which extent the target at day 0 (absence in 25 g) can be extended. If this criterion is reached, it can be avoided that the limit of 2 log CFU/g is exceeded at the end of shelf-life. Partners: KU Leuven Dept. of Chemical Engineering, Chemical & Biochemical Process Technology & Control (BioTeC) UGent Dept. of Food Safety & Food Quality, Laboratory of Food Microbiology & Food Preservation (LFMFP) 37
PARIS, SEPTEMBER 16-20, 2013 CO-ORGANIZED by: supported by: ASTR A sponsored by: