Parametrization of microbial survival models under UVC exposure

Abstract

This work presents a unified framework for the parametrization and comparison of microbial survival models under UVC exposure. Four dose-response models - the single-target, multi-target, linear-quadratic, and two-stage decay models - are analyzed with respect to their mathematical structure, parameter identifiability, and biological interpretability, while ensuring physically meaningful parameter estimates. To address the limited size and lack of replication in published datasets, parameter uncertainty is quantified using a parametric bootstrap approach under multiplicative dose uncertainty. Model comparison combines goodness-of-fit in logarithmic survival space, the Akaike Information Criterion corrected for small samples, and identifiability considerations. Application to data for 32 microorganisms shows that no single model is universally optimal, highlighting that reliable model selection requires combining statistical performance with physical and biological consistency, and providing a robust basis for UVC survival analysis and disinfection modeling.