Single Au@Pt bimetallic nanoparticle collision electrochemistry: Deep learning-assisted prediction and ultrasensitive...

Single-particle collision electrochemical technology (SPCE) shows significant potential for nanoscale sensing, owing to its high spatiotemporal resolution and label-free characteristics. Nonetheless, its application in real biological samples faces challenges such as limited selectivity and complications in quantitative signal analysis due to complex matrix interference. This study introduces a novel detection strategy that utilizes electrocatalytic signal switching of gold-platinum nanoparticles (Au@Pt NPs) in conjunction with deep learning algorithms, enabling highly sensitive and specific quantitative analysis of matrix metalloproteinase 9 (MMP-9) at the single-particle level. The engineered Au@Pt NPs exhibit a synergistic electrocatalytic effect on the oxidation of hydrazine (N₂H₄) by using a gold ultramicroelectrode (Au UME), generating a stepped current response with the signal intensity approximately 2.5 times greater than that of platinum nanoparticles (Pt NPs). By functionalizing the nanoparticle surface with MMP-9 specific DNA aptamers, we create a steric hindrance structure that effectively inhibits the nanoparticles' catalytic activity for N₂H₄ oxidation. In the presence of MMP-9, the specific binding interaction re-exposes the catalytic sites, restoring the electrochemical collision signals. To map the complex signals accurately to the concentration of the target analyte, a deep learning model was employed to automatically extract features and conduct regression analysis on the raw i-t trajectory, utilizing a dataset of collision signals triggered by varying concentrations of MMP-9. This study presents an innovative approach that integrates advanced data analysis with single-particle electrochemistry, enhancing the ultrasensitive identification of tumor markers.