Machine learning-enhanced electrochemical sensing platform for simultaneous detection of multiple antibiotics in food...

The increasing prevalence of antibiotic residues in food presents a significant threat to global public health. To address this critical concern, we developed a novel artificial neural network (ANN)-enhanced electrochemical sensor based on ZIF-8/MnMoO₄/MWCNTs-modified glassy carbon electrode (GCE) with high sensitivity for the simultaneous detection of chloramphenicol (CAP), nitrofurazone (NFZ), and metronidazole (MNZ) in complex food matrices. The sensor exhibited excellent analytical performance, featuring a wide linear detection range of 0.005-00.00 μM and ultralow detection limits of 0.08 μM (CAP), 0.29 μM (NFZ), and 0.23 μM (MNZ), respectively. Notably, high selectivity was achieved with well-separated oxidation peaks, and the interference from coexisting substances was negligible (experiments conducted in milk and honey samples yielded reliable results with recoveries ranging from 90.0 % to 110.0 % and relative standard deviations (RSD) below 4.07 %, confirming the applicability of the sensor in the detection of real food matrices. To further enhance analytical performance, machine learning (ML) models were employed to optimize detection parameters and improve prediction accuracy. Specifically, a backpropagation artificial neural network (BP-ANN) was utilized for signal interpretation, achieving a prediction accuracy exceeding 93 % and significantly reducing regression errors. Collectively, this intelligent electrochemical sensor demonstrates robust performance in real food sample analysis, highlighting its great potential as a practical platform for future food safety monitoring of multiple antibiotic residues.