Workshops
Workshop 1
Title of the Workshop
Exploiting the power of the harmonized knowledge exchange format FSKX
Organized by:
Matthias Filter (Contact), German Federal Institute for Risk Assessment (BfR)
Laurent Guillier, French Agency for Food, Environmental and Occupational Health & Safety (ANSES)
Yvonne Mensching, German Federal Institute for Risk Assessment (BfR)
Thomas Schüler, German Federal Institute for Risk Assessment (BfR)
Maarten Nauta, Statens Serum Institut (SSI)
Fernando Perez Rodriguez, University of Cordoba (UCO)
Aricia Possas, University of Córdoba (UCO)
Racem Ben Romdhane, German Federal Institute for Risk Assessment (BfR)
Justification and Objective
The objective of the workshop is to introduce the Food Safety Knowledge eXchange (FSKX) format and to showcase how model creators and model users can benefit from it. Several software tools already support the creation, joining, execution and sharing of FSKX models. Attendees will learn about the FSKX format, the underlying model exchange strategy and the different software solutions like online model repositories, web-based services to generate, edit, join or execute FSKX model files as well as a peer-reviewed journal allowing to import FSKX files for the efficient creation of modelrelated publications (https://fesmj.pensoft.net/). Participants will perform hands-on exercises with the RAKIP-Web model repository and MicroHibro where FSKX formatted models are available for download and execution.
The workshop is supported by the Risk Assessment Knowledge Integration Platform
(RAKIP) Initiative which is driven by several European risk assessment agencies and universities. The RAKIP Initiative promotes the development of resources that allow efficient exchange and re-use of models, data and simulation results in One Health related sectors. With that it addresses the need of researchers to find, apply and customize existing predictive models irrespective of the programming language or software used. The RAKIP Initiative also maintains the so-called Generic Metadata Schema that is used by FSKX to annotate models, data and simulations in a harmonized way. RAKIP Initiative members further develop open source software resources that can be integrated into 3rd party software tools.
Workshop attendees get the opportunity to perform their own hands–on exercises with online model repositories and to create their own FSKX model files. After the workshop, participants are familiar with the RAKIP Initiative, the FSKX format, and the opportunities emerging from adopting the format for own research and modelling results.
Programme of the Workshop (3 hours)
Introduction of presenters, housekeeping & moderation (10 min)
Y. Mensching, (BfR)
FSKX format – a format for models, data and simulation results (20 min)
M. Filter, BfR
Live Demo RAKIP-Web Model Repository (20 min)
L. Guillier (ANSES)
Hands-on exercise – Model Execution @RAKIP-Web (40 min)
R. Ben Romdhane (BfR)
Break (15 min)
Live Demo – Model Joining (20 min)
M. Nauta (SSI), T. Schüler (BfR)
Live Demo FESMJ – FSKX model paper (20 min)
T. Schüler, Y. Mensching, R. Ben Romdhane, M. Filter (BfR)
Outlook: Connecting Distributed Model Repositories (20 min)
F. Perez Rodriguez, A. Possas (UCO)
Q&A (15 min)
M. Filter (BfR)
End of the workshop
Requisites for Attendees
To participate in the hands-on exercise participants should bring their laptop with WLAN connectivity.
Workshop 2
Title of Workshop
Bayesian methods for microbiological data and risk assessment
Organized by:
Dr. Jukka Ranta (Contact), Finnish Food Authority
Dr. Anne Thebault, French Agency for Food, Environmental and Occupational Health & Safety (ANSES)
Dr. Kento Koyama, Hokkaido University
Dr. Hiroki Abe, Hokkaido University
Justification and Objective
The overall objective of the three-hour workshop is to give a theoretical introduction and practical examples of using Bayesian methods for bacterial modeling in food safety. These are based on MCMC simulations, utilizing Bayesian software such as OpenBUGS/JAGS/Stan in tandem with R. The practical objective of the workshop is to familiarize with the theoretical concepts and how they can be applied in microbiological problems involving uncertainty and variability. A central concept is probabilistic inference, which is conducted to obtain probability statements of the unknown model parameters of interest, conditionally on the observed data. This means inferring holistically all unknowns jointly together in one step, in the form of a multidimensional uncertainty distribution, which is known as the posterior distribution. Computation of such distributions requires numerical methods for which special software is usually needed. Since the distribution represents the combined uncertainty of several parameters, conditionally on data, it can be used for investigating how well actually the parameters are identified from data. That is, what, and in what accuracy, could be learned from data and whether it is different from the prior knowledge. The concept of prior distribution will be explicit in each example. The Probabilistic representation of uncertainty can further be employed to quantify uncertainty in other derived quantities and distributions which depend on those unknown parameters, e.g. to describe uncertainty of quantiles of variability distributions. However, the quality of fitting of Bayesian models requires specific controls. Examples include censored concentration distributions, growth/inactivation models and dose-response models.
Participants are expected to bring their laptops and have a Google account for using Google Colaboratory.
Outline
Part I: introduction to the theory (1/2 hour) (likelihoods, priors, principle of MCMC, DAG, posterior controls & quality of fitting)
Part II: introduction to the software (1/2 hour) (BUGS/JAGS/Stan, and the associated tools for using them, such as e.g. R2OpenBUGS & brms for R, PyStan & Google Colab for Python).
Part III: practical examples (2 hours). Some concrete examples could be the following:
Workshop 3
Title of Workshop
Introduction to Predictive Microbiology
Contact Person
Dr. Lihan Huang, Research Leader, Residue Chemistry and Predictive Microbiology Research Unit, Eastern Regional Research Center, Agricultural Research Service, U.S. Department of Agriculture Research Service.
600 E. Mermaid Lane, Wyndmoor, PA 19038. USA.
Email: lihan.huang@usda.gov
Justification and Objective
In recent years, significant progress has been made in predictive modeling research and application. Many predictive models, tools, and databases have been developed for data analysis, model development, and risk assessment, and are available to the industry, academia, international organizations, and governments around the world. Many of these technical resources and application tools, available online or on desktop, can provide a fast and reliable decision-making process for food safety and quality in the industry. Typical applications of predictive microbiology may include prediction of microbial behavior during food processes and storage conditions, shelf-life prediction, performance and validation of sampling plans, and quantitative risk assessments.
This workshop attempts to summarize, present, and discuss the most recent developments, demonstrate both fundamental and applied aspects of predictive microbiology, and will also introduce the most up-to-date one-step dynamic analysis in predictive modeling. This workshop will use the USDA Integrated Pathogen Modeling Program (USDA-IPMP) to demonstrate the basic concepts of predictive microbiology and introduce one-step kinetic analysis using the USDA IPMP-Global Fit. The aim of this workshop is to introduce basic concepts, experimental design, data analysis, and practical use of these computing tools to develop accurate predictive models more effectively and efficiently.
Targeted audience: Students, scientists, and engineers who are interested in developing predictive models for microbial shelf-life prediction and risk assessment.
Proposed Speakers and Agenda
(total 2 h 40 min + 20 min coffee break)
1. Introduction to predictive microbiology, model development, and applications
2. Primary growth and survival models and data analysis
3. Secondary models and data analysis
4. One-step kinetic analysis
5. Tertiary models and application
Prerequisites:
For basic concepts, none.
Workshop 4
Title of the Workshop
Meta-Regression in Food Microbiological Safety
Organized by:
Ursula Gonzales-Barron, Instituto Politécnico de Bragança, Portugal
Vasco Cadavez, Instituto Politécnico de Bragança, Portugal
Justification and Objective
“Meta-Regression in Food Microbiological Safety” is an ICPMF12 pre-conference workshop whose objective is to introduce meta-regression modelling methods that can be used to summarise and contrast available knowledge on various food microbial safety issues. The relevance of meta-regression methods will be demonstrated by reviewing useful applications such as the estimation of the overall log reduction of a pathogen for a given risk-mitigation treatment, and the estimation of overall microbial kinetic parameters (such as growth rate) with basis on predictive microbiology equations. In this 2.5-hour workshop, an overview of the general methodology of meta-analytical regressions fitted as multilevel models will be provided. The workshop is structured in brief theoretical sessions, lectured by Dr. Ursula Gonzales-Barron, followed by hands-on demo sessions using the R software, lectured by Dr. Vasco Cadavez. The participants are expected to have a general knowledge on microbiology and statistics, and should bring their own laptops with R (http://www.r-project.org) and RStudio (http://www.rstudio.org) installed.
Maximum number of participants
30
Outline
(total 2 h 30 min + 20 min coffee break)
1. Brief introduction to meta-analysis
2. Meta-regression modelling
3. Application 1: Use of meta-regression for modelling log-reduction effectiveness of a microbiological risk-mitigation treatment
4. Application 2: Use of meta-regression for extracting microbial kinetic parameters
Prerequisites: Linear regression