Nanogap-Assisted SERS/PCR Biosensor Coupled Machine Learning for the Direct Sensing of <i>Staphylococcus aureus</i> in Food

Abstract

<i>Staphylococcus aureus</i> (<i>S. aureus</i>) is the primary risk factor in food safety. Herein, a nanogap-assisted surface-enhanced Raman scattering/polymerase chain reaction (SERS/PCR) biosensor coupled with a machine-learning tool was developed for the direct and specific sensing of S. aureus in milk. The specific <i>nuc</i> gene (<i>nuc</i> T) from <i>S. aureus</i> was initially amplified through PCR and subsequently captured via the nanogap effect of I^- and Mg²⁺-mediated bimetallic gold and silver nanoflowers (Au/Ag FL@I^--Mg²⁺). These nanogaps generate hotspots for the direct signal amplification of enclosed <i>nuc</i> T. Subsequently, machine-learning tools were used to comparatively analyze the collected SERS signals. The bootstrapping soft shrinkage-partial least-squares method exhibited superior performance (root mean-square error of prediction: 0.437, prediction set correlation coefficient: 0.967). This study demonstrated a novel label-free strategy for specifically detecting <i>S. aureus</i>. The strategy could be advanced to serve as a platform for application to other types of foodborne pathogenic bacteria by engineering a suitable specific primer.