An Intelligent Machine-Learning-Driven Metalloporphyrin Covalent Organic Framework Sensor Array for Pesticide Discrimination

Pesticides are an indispensable component of agricultural and food management to environmental protection and human health. Unquestionably, rapid identification of various pesticide residues is an urgent issue requiring resolution yet remains challenging. This study introduces a colorimetric sensor array based on metalloporphyrin covalent organic frameworks (MCOFs) for the rapid and accurate discrimination of multiple pesticides. The array utilizes three distinct MCOF nanozymes as catalytic sites, which enhanced discrimination activity by 2.4-4.5-fold compared to metal-free COFs. Aided by Linear Discriminant Analysis (LDA), the array successfully differentiated five representative pesticides within a 10-100 nM range, achieving limits of detection (LODs) as low as 2.19 nM. Furthermore, by integrating a Random Forest (RF) algorithm, the platform demonstrated high performance in complex matrices. It achieved 94.1% accuracy in identifying mixed pesticide residues in river water and over 95% accuracy for classifying pesticides in tomato samples based on their functional groups. This work would provide a simple, convenient, and promising approach for distinguishing, hierarchically screening to predict multiple pesticide residues in environmental and food monitoring applications.