Development and Application of search tools for mining and extrapolating currently available data for predictive model development

Fermented and smoked foods accounted for five percent of total meat consumption in Europe. With domestic demand increasing and a wide range of products available in major retail outlets, these foods were produced by both multinational companies and artisan micro food businesses. Regardless of size, all companies required predictive models for determining shelf-life and safety, as these foods were susceptible to microbial spoilage and could carry pathogens like Listeria monocytogenes.

This project developed and applied search tools to mine and extrapolate existing relevant data for predictive model development. The research team conducted broth-based studies to address data gaps, focusing on the impact of key parameters—such as pH, temperature, water activity, and preservatives—on the growth of spoilage bacteria like *Brochothrix thermosphacta*, *Photobacterium phosphoreum*, *Shewanella putrefaciens*, *Pseudomonas fluorescens*, and *Enterobacteriaceae*, as well as pathogenic bacteria like *Listeria monocytogenes* and *Clostridium sporogenes*.

Growth curves for each bacterium were individually fitted using the Gompertz function through non-linear regression. Descriptors such as lag phase duration, exponential growth rate, and generation time were calculated, with polynomial models developed to relate these factors to the key parameters. These models were validated in representative fermented and smoked foods, and packaging conditions were optimized accordingly.

The final task focused on dissemination and technology transfer, delivering spoilage and safety data files, optimized packaging technology, and the predictive models.