Food Safety Predictive Modeling

Translating science into practice has benefits for food producers seeking to improve food safety for consumers. Applying a tool like predictive microbiology to a commercial environment could help producers plan work processes effectively with consumer safety in mind. Output from predictive models could predict shelf life and identify control points, for example. However, implementation of such schemes is a challenge. In a recent paper, Plaza- Rodríguez et al. (2015) describe how they sought to overcome these obstacles by setting up predictive food safety model repositories focused on making previously published research data on Salmonella in beef available to producers¹.

Plaza- Rodríguez et al. describe the project as community-driven, with the intention to share data from scientific literature on Salmonella growth and survival with food producers and regulators. The authors believe that enabling access to this information means it could be more widely applied to all stages of the food production and processing workflow. By developing a model that could predict bacterial growth and behavior, food producers and consumers alike could benefit from improved product safety as the information could be more readily applied to HACCP planning and other management protocols.

The problem, however, is that this information is either not widely available to industry, or that it is in a form that is not accessible. Furthermore, predictive models may only be accessible within certain software formats and users may not be able to upload data depending on the file types handled. Even though the data contained within scientific literature is extremely useful, it may not successfully travel into the commercial world because of these obstacles.

The researchers used PMM-Lab v.1.06 to create the framework for the predictive model, choosing this software because it is open source and therefore fosters collaboration among researchers. Plaza-Rodríguez et al. also felt the software brought additional benefits to the project in terms of transparency in data processing that could be seen by all collaborators. It also allows users to bring in additional models into the repository scheme. The team also examined using Systems Biology Markup Language (SBML) as a common language within the data exchange.

The researchers used three different approaches to create predictive food safety models, referencing work from previously published studies as source data;

Implementation using only data available in scientific literature
Implementation using existing curve-fitting algorithms and predictive models
Implementation using experimental data generated within the predictive model using PMM-Lab itself

Once created, the researchers checked validity of the results generated using Pathogen Modeling Program (PMP), examining generated growth curves, inactivation curves and other output for similarities. The team also checked suitability for interaction with other software tools and data sources. They found that each approach offered benefits to the end user dependent on the type and completeness of the data available for model creation. Furthermore, with certain formatting requirements, SBML is an appropriate language to use within the data exchange environment.

In conclusion, Plaza-Rodríguez et al. believe that developing food safety model repositories is beneficial to the food industry, and that using open source tools like PMM-Lab in conjunction with a defined data information exchange format is preferred. By using open access and standardized formats, the researchers believe that a web-based public format is highly possible and practical.

For further discussion on microbial food safety visit our food and beverage community

You may also be interested in our previous posts on Controlling Salmonella Contamination and Modeling Contamination in Poultry Meat

Reference

1. Plaza-Rodríguez, C. et al. (2015) “A strategy to establish Food Safety Model Repositories“, International Journal of Food Microbiology 204 (pp.81–90)