Machine Learning-Enhanced Nanozyme Sensor Array for Accurate Multiple Quinolone Antibiotics Recognition

The overuse of quinolone antibiotics (QNs) seriously endangers human health and the ecological environment. In this work, a copper dihydroxosulfate (Cu2(OH)2SO4) nanosheet exhibiting notable peroxidase-like (POD) and laccase-like (LAC) activities has been developed in basic deep eutectic solvents (DES). The unique physicochemical properties of QNs allow them to enhance the POD activity of Cu2(OH)2SO4, and with the extension of reaction time, this enhancement gradually intensifies. Conversely, when QNs are introduced into the LAC reaction system of Cu2(OH)2SO4, they significantly inhibit its LAC activity, with the degree of inhibition growing increasingly evident as the reaction time increases. A nanozyme sensing array has been developed via reaction dynamics to identify eight QNs. This method cleverly achieves self-calibration through two reverse signals, further improving the sensing performance of the sensor array. Moreover, through the optimization of various machine learning (ML), the precision of the concentration-independent recognition model built upon this array has been enhanced from 39.08% to 91.95%. This improvement is advantageous for the identification of unknown samples within actual samples. This work carries significant implications for enhancing the discrimination of QNs in complex samples.