Jinxin Liu, Marti Z Hua, Xinyu Yan, Luyao Ma, Shenmiao Li, Yang Wang, Tian Yang, Yihan He, Michael E Konkel, Greta Gölz, Thomas Alter, Jinsong Feng, Qian Liu, Xiaonan Lu
This fully automated and low-cost system minimizes human error and workload, providing a scalable sample-to-answer solution for AMR surveillance in global agri-food system.
INTRODUCTION: Antimicrobial resistance (AMR) threatens food safety across the farm-to-fork continuum. Real-time surveillance is crucial to mitigate its global escalation, yet conventional antimicrobial susceptibility testing (AST) remains slow, labor-intensive, and impractical for large-scale monitoring.
OBJECTIVES: We developed an Artificial Intelligence of Things (AIoT)-integrated multiplex microfluidic platform enabling automated AMR surveillance ofpathogens in food supply chain.
RESULTS: Each node combines a single-board AIoT controller (Orange Pi 5B), portable incubator, colorimetric microfluidic chips, and environmental sensors, reducing costs by 98% compared with standard AST. A lightweight YOLO model embedded in the controller achieved > 99% accuracy in identifying bacterial growth and inhibition under antibiotic pressure, showing 96% and 95% agreement with standard results forSalmonella andCampylobacter, respectively. Data are synchronized to a cloud server for real-time aggregation and early resistance warning.
CONCLUSION: This fully automated and low-cost system minimizes human error and workload, providing a scalable sample-to-answer solution for AMR surveillance in global agri-food system.