Machine Learning-Driven Sensor Array Based on a DNA-Programmed Pt Nanozyme with Enhanced Dual-Enzyme-like Activities for Discrimination of Multiple Sulfur-Containing Species

Abstract

Nanozyme sensor arrays have garnered significant attention in the sensing field, whereas their practical application remains limited due to the issues of multiple independent sensing elements, concentration interference effects, and challenges in real-world implementation. Herein, we propose a machine-learning-empowered sensor array based on a single nanozyme with dual-enzyme-like activities for the simultaneous discrimination of multiple sulfur-containing species (S2–, S2O32–, S2O82–, SO32–, and SO42–). Hence, a DNA-modulating strategy to fabricate Pt nanozymes with improved peroxidase-like and laccase-like activity was proposed, and the results indicated that the A10@Pt nanozyme showed higher peroxidase-like and laccase-like activity than other Pt nanozymes, which were 5.5- and 3.5-fold than that of pure Pt nanozymes, respectively. As a proof of concept, a single A10@Pt nanozyme with dual-enzyme-like activities was utilized as a sensor element to develop a novel sensor array for the discrimination of five sulfur-containing species (S2–, S2O32–, S2O82–, SO32–, and SO42–). Additionally, a machine-learning algorithm was used to construct a stepwise prediction model for integrating the proposed nanozyme sensor array, enabling accurate identification and prediction of five sulfur-containing species, which was successfully validated in actual water samples (canal water and Yangtze River water). Thus, our work provided a robust, rapid, and intelligent detection platform for sulfur-containing species in environmental monitoring, enabling a valid assessment of sulfur pollution and its ecological impacts.