Machine Learning-Assisted Concentration-Independent Recognition of Neonicotinoids Based on Multienzyme-like Activities of FeCu Dual-Atom Nanozyme

Residues of neonicotinoid insecticides (NEOs) pose serious threats to ecological systems and human health. Conventional nanozyme sensors often suffer from limited catalytic diversity and concentration-dependent response mechanisms, which lead to signal homogenization and cross-concentration misclassification. To address these limitations, we developed a Fe-Cu dual-atom nanozyme (FeCu DAzyme) exhibiting triple-enzyme activities: oxidase (OXD), peroxidase (POD), and laccase (LAC). The synergistic effects between Fe-Cu dual-atom sites significantly enhanced catalytic efficiency, while their specific coordination with NEO functional groups enabled distinct inhibition responses across different concentration levels. Leveraging this property, we constructed a FeCu DAzyme-based colorimetric sensor array that captures real-time inhibition kinetics of OXD/POD/LAC activities, generating unique multidimensional response patterns. Through integration with a machine learning classifier, these patterns enabled accurate pesticide identification independent of absolute concentration values. The sensor array achieved 92.50% accuracy in discriminating five NEO structural analogs across a concentration range of 0.1-50 μg/mL, demonstrating excellent concentration-independent identification capability. Notably, the practical utility of this platform was successfully validated through the high-accuracy identification of NEOs in spiked real-world samples, including lake water and agricultural products. This work established a promising paradigm for rapid NEO identification, which is critical for ensuring agricultural product safety.