Artificial intelligence-assisted colorimetric sensor array based on supramolecular self-assembled nanozymes for visual monitoring of pesticide residues

In this study, we designed a nanozyme-based colorimetric sensor array (CSA) and utilized deep learning technology to achieve intelligent detection of multiple pesticide residues. Unlike other sensor arrays, this one employ Cu-AC nanozymes self-assembled from different amino acids (L-leucine, L-isoleucine, and L-phenylalanine) with copper ions (denoted as Cu-Leu, Cu-Ile, and Cu-Phe) as the sensing units. The laccase activity of the nanozymes can be easily regulated through the interaction between amino acids and copper ions. The simple and efficient preparation process significantly reduces manufacturing costs and enhances the controllability of enzyme activity. The three nanozymes, combined with the substrate (2,4-dichlorophenol, 2,4-DP) and the chromogenic agent (4-aminoantipyrine, 4-AP), form a colorimetric sensor array for linear discriminant analysis (LDA) and hierarchical cluster analysis (HCA) of seven pesticide residue. This ultimately achieves highly sensitive detection with a detection limit of less than 0.0012 μM. Additionally, by using the deep learning YOLOv8 algorithm to visually analyze and train the original LDA plots, the sensor array can automatically classify and detect the different pesticide residues. The results show that the model's mean average precision (mAP) on the validation dataset reaches 0.99, and its average confidence on the testing dataset can reach 0.93 (on a scale of 0 to 1). The YOLOv8-assisted approach reduces detection time and improves accuracy, providing robust technical support for pesticide residue detection in complex environments.