Abstract
Microbial growth and inhibition are complex biological processes influenced by diverse environmental and chemical factors, posing challenges for accurate modelling and prediction. Traditional mechanistic models often struggle to capture the nonlinear and multidimensional interactions inherent in real-world food systems, especially when multiple environmental variables and inhibitors are involved. This study presents the development of a novel, dynamic software platform that integrates classical predictive microbiology models-including both one-step and two-step frameworks-with advanced machine learning (ML) methods such as Support Vector Regression, Random Forest Regression, and Gaussian Process Regression. Uniquely, this platform enables direct comparisons between two-step and one-step modelling approaches across four widely used growth models (modified Gompertz, Logistic, Baranyi, and Huang) and three inhibition models (Log-Linear, Log-Linear + Tail, and Weibull), offering unprecedented flexibility for model evaluation and selection. Furthermore, the platform incorporates ML-based modelling for both microbial growth and inhibition, expanding predictive capabilities beyond traditional parametric frameworks. Validation against experimental and literature datasets demonstrated the platform's high predictive accuracy and robustness, with machine learning models, particularly Gaussian Process Regression and Random Forest Regression, outperforming classical models. This versatile platform provides a powerful, data-driven decision-support tool for researchers, industry professionals, and regulatory bodies in areas such as food safety management, shelf-life estimation, antimicrobial testing, and environmental monitoring. Future work will focus on further optimization, integration with large public microbial databases, and expanding applications in emerging fields of predictive microbiology.