Transferable artificial intelligence for viable but non-culturable foodborne pathogen detection: opportunities, challenges, and future directions
Abstract
The viable but non-culturable (VBNC) state enables foodborne pathogens to evade culture-based detection while retaining virulence and resuscitation capacity. This minireview examines the intersection of artificial intelligence and VBNC research. We survey transferable artificial intelligence (AI) methods from adjacent microbiology domains, identify the scarcity of direct AI-VBNC studies and its root causes, and delineate critical challenges, including data scarcity, standardization, and model interpretability. We also construct a transferability framework mapping existing AI tools to specific VBNC research needs with feasibility assessment and propose a roadmap for multimodal integration, benchmark data sets, and explainable AI to advance the field.