Dual-mode integrated electrochemical sensing of E. coli in real matrices enabled by machine learning.

For accurate detection of microbial indicators in spiked food matrices, a dual-mode electrochemical biosensor based on a magnetic nanoparticle (MNPs)-aptamer (Apt) complex was developed for highly sensitive quantification of Escherichia coli (E. coli). Fe3O4@Au nanoparticles were synthesized and functionalized with Apt via Au-S bonds to form the Fe3O4@Au@Apt complex. In the "signal-on" mode, the charge transfer resistance of the [Fe(CN)6]3-/4- probe increased as measured by electrochemical impedance spectroscopy (EIS); in the "signal-off" mode, the oxidation peak current of methylene blue (MB) decreased using differential pulse voltammetry (DPV). By integrating features from both EIS and DPV responses, 11 concentration-related features were extracted. A genetic algorithm (GA) was employed to optimize the hyperparameters of an XGBoost model for accurate prediction of E. coli concentrations in real samples. This dual-mode strategy integrates the complementary strengths of EIS and DPV, achieving a linear detection range from 101 to 107 CFU/mL with high recovery rates in real samples. The approach offers a robust and reliable tool for food safety and environmental monitoring.