Machine learning-assisted universal PGM sensing platform: Recognition-detection separation via DNAzyme-nanozyme triplex switch.

Constructing a universal, accurate personal glucose meter (PGM)-based detection strategy via separating recognition and detection to eliminate matrix interference, and a versatile signal transduction method without glycosidases, can enhance PGM detection universality and accuracy, expanding its point-of-care testing (POCT) applications beyond blood glucose detection. Herein, we develop a highly sensitive PGM-based biosensor for ochratoxin A (OTA) detection by integrating a triplex molecular switch (THMS) with DNAzyme-mediated signal amplification using an Au@Pt nanozyme. The THMS, assembled from an aptamer and a Pb2+-dependent DNAzyme, specifically recognizes OTA, triggering THMS dissociation and DNAzyme release. The liberated DNAzyme cleaves the Au@Pt-SingleDNA-MB signal tag, releasing Au@Pt nanoparticles for PGM readout via glucose oxidation. Conventional four-parameter logistic (4 PL) regression was used to establish the calibration curve, which showed a linear range of 0-0.05 ng/mL and a detection limit of 0.008 ng/mL; this detection limit is 10-fold lower than that of Au NPs-based methods. Additionally, machine learning was employed to further analyze the performance of the biosensor. Recovery experiments in corn and wine samples gave acceptable recoveries of 89.09%-116.66% and 83.90%-109.09%, demonstrating the practical applicability of the biosensor. Overall, this study provides a portable and sensitive POCT platform for the on-site detection of mycotoxins.