Extreme gradient boosting-enhanced molecularly imprinted fluorescence nanosensor for rapid identification and visual detection of deltamethrin in seawater and aquatic products

Abstract

Deltamethrin (DEL), a synthetic pyrethroid insecticide, may have long-term adverse effects on the aquatic environment. Extreme Gradient Boosting (XGBoost) is a classic ensemble boosting algorithm framework known for its high training efficiency, strong prediction performance, and versatile applications. In this study, we successfully developed a highly sensitive molecularly imprinted fluorescence nanosensor (MF-sensor) for DEL determination, enhanced by the XGBoost algorithm. The MF-sensor was fabricated by grafting blue and red CdSe/ZnS quantum dots onto a molecularly imprinted silica layer. The MF-sensor consisted of blue-molecularly imprinted polymer (MIP)-quantum dots (QDs) and r-MIP-QDs, which selectively detected DEL, and unmodified green CdSe QDs, which were not selective for DEL. Under optimized conditions, we observed an excellent linear relationship between the I526/(I450 +I630) ratio and DEL concentration ranging from 0.01 mg/L to 40.0 mg/L (R2 = 0.9944). The limit of detection was determined to be 1.34 µg/L. The recoveries in actual samples ranged from 98.5% to 109.0%, with a relative standard deviation (RSD) below 7.5%. Additionally, we utilized the XGBoost algorithm to establish a DEL prediction model, achieving an accuracy of 95.7%. The recoveries ranged from 96.0% to 102.4%, with a RSD below 5.5%. Overall, the proposed MF-sensor, enhanced by the XGBoost algorithm, was successfully utilized for the detection of DEL in environmental and aquatic products.