Active capture-directed bimetallic nanosubstrate for enhanced SERS detection of Staphylococcus aureus by combining strand exchange amplification and wavelength-selective machine learning.

Staphylococcus aureus (S. aureus) is the leading risk factor for food safety and human health. Herein, a novel wavelength-selective machine learning -driven adaptive strand exchange amplification (SEA)/SERS biosensor was developed for rapid detection of S. aureus. The study operates via three innovative routes: i) the exceptional specificity triggered through SEA of the nuc target gene (nuc T') from S. aureus, ii) anodic aluminum oxide filter membrane-supported gold and silver bimetallic nanoflowers modified with 4-ATP (Au/Ag FL@AAO-4-ATP) were prepared as SERS nanosubstrate for actively capturing the nuc T' through a nucleophilic addition reaction, and for SERS signal enhancement, and finally iii) the integration of a wavelength-selective machine learning tool for further refinement and accuracy of the S. aureus detection process. The proposed wavelength-selective machine learning-driven adaptive Au/Ag FL@AAO-4-ATP nanosubstrate administers prediction performance for the quantitative detection of S. aureus using interval combined population analysis-partial least squares (ICPA-PLS) with root mean square error of prediction and residual predictive deviation values of 0.9626 and 3.56, respectively. The effectiveness of the proposed ICPA-PLS method in real milk samples was validated by a standard fluorescent quantitative PCR method in terms of t-test with no significant differences at P = 0.05. This study offers a new avenue for rapid and straightforward detection of S. aureus, focusing on key genes. The proposed SERS/SEA/machine learning integrated platform can be adapted to other bacterial species via engineering appropriate amplification primers, thus, inspiring potential applications in food safety and biomedical research.