AI-Assisted Electrochemical Immunosensing for Matrix-Aware Detection of Aflatoxin M1 and Atrazine in Food Matrices

Food contamination by Aflatoxin M1 and Atrazine remains a critical food-safety concern, requiring sensitive detection methods that can operate reliably in complex matrices. Here, we report an AI-assisted antibody-functionalized electrochemical sensing platform for the detection and classification of Aflatoxin M1 and Atrazine across corn, corn flour, and protein matrices. The sensor used analyte-specific antibodies immobilized on an electrochemical electrode surface, where target binding produced measurable changes in the interfacial electrochemical response. Sensor performance was evaluated using cyclic voltammetry, coulometry, and electrochemical impedance spectroscopy (EIS), with EIS providing strong frequency-dependent signatures for concentration-dependent analysis. Spike-and-recovery studies further demonstrated the applicability of the platform in food-matrix conditions. To improve interpretation of complex electrochemical signals, full-spectrum EIS features were integrated with machine learning models for concentration-level classification into low, mid, and high groups. The AI workflow achieved an overall classification accuracy of 93.33%, with 96.67% specificity, 93.44% PPV, 96.66% NPV, and 0.982 AUC for Atrazine, and 96.70% specificity, 93.38% PPV, 96.67% NPV, and 0.987 AUC for Aflatoxin M1. In addition, analyte classification between Aflatoxin M1 and Atrazine reached 97.4% accuracy and 0.994 ROC-AUC. Overall, this work demonstrates a matrix-aware electrochemical immunosensing strategy enhanced by AI-based signal interpretation for food contaminant detection.