Machine Learning–Enhanced Detection of Antimicrobial Resistance in Foodborne Pathogens: Advances, Challenges, and Future Directions

Antimicrobial resistance (AMR) in foodborne pathogens poses a growing global threat to food safety, public health, and international trade. Conventional detection approaches, including culture-based, molecular, and phenotypic methods, remain essential for AMR surveillance; however, their limitations in speed, scalability, and integrative capacity hinder timely risk assessment in increasingly complex food systems. Recent advances in machine learning (ML) provide an analytical complement, particularly for integrating heterogeneous genomic, proteomic, spectroscopic, and environmental data. This review synthesizes ML-enhanced AMR detection in foodborne pathogens, spanning resistance biology and transmission in the food chain, conventional diagnostics, and ML-driven predictive models. Advances in ensemble learning, deep neural architectures, and hybrid approaches are discussed alongside their integration with biosensing, spectroscopic, and microfluidic technologies for rapid and data-driven detection. The review also highlights key challenges limiting real-world deployment, including data limitations, interpretability constraints, infrastructure requirements, and regulatory barriers. Finally, future prospects for integrating ML into food safety monitoring systems are explored within a One Health framework. Emphasis is placed on data harmonization, explainable and interoperable models, and coordinated governance as prerequisites for translating ML-based AMR detection into reliable, scalable, and policy-aligned food safety surveillance.