From data to models and predictions in food microbiology

Abstract

This paper emphasizes the importance of structured databases, visualization techniques, statistics, and mathematical models as milestones when developing predictive models of bacterial responses to food environments. Predictions generated by such models are vital in decision-making on food safety and quality issues. The paper suggests that while refinements, such as reparameterization, rescaling, and fine-tuning smoothness-characterizing parameters, are useful for numerical/statistical point of view, the result should not be considered as new models. It is proposed that novel predictive models should be linked to those widely accepted in related disciplines, such as biotechnology, systems biology, or biochemistry.