Integrated microneedle patch and machine learning in time-resolved fluorescent immunochromatography test strip for accurate and rapid detection of Staphylococcus aureus enterotoxin B in salmon.

Abstract

Staphylococcus aureus enterotoxin B (SEB) is a leading cause of foodborne illness. In this study, we developed a time-resolved fluorescent immunochromatography test strip (TRFITS) combined with a hydrogel microneedle patch (HMNP) and machine learning (ML) for the rapid detection of SEB in salmon. The developed HMNP, fabricated from methacrylated gelatin, can automatically extract 37 mg of interstitial fluid within 6 min. We then conducted TRFITS analysis on salmon samples to detect SEB and ML algorithms so as to predict its concentration. This method enabled efficient sampling and accurate quantification of SEB within 32 min, with a limit of detection (LOD) of 4.63 ng/mL. In addition, ML-assisted TRFITS applied to salmon demonstrated that the YOLOv11 model achieved a prediction accuracy of 91.3% and a lower LOD of 2 ng/mL. These findings highlight a novel approach toward on-site SEB detection with significant implications for food safety monitoring and point-of-care diagnostic applications.