A deep learning-based Cu SAzymes sensor ：Preparation and their application in sweat monitoring

Abstract: Using Cu-ZIF-8 as a precursor, copper single-atom nanozymes (Cu SAzymes) were successfully prepared through a high-temperature pyrolysis strategy. AC-HAADF-STEM characterization confirmed that copper elements were dispersed at the atomic level within the material. Enzyme-catalyzed kinetic studies showed that the Michaelis constant Km for catalyzing H₂O₂ was 0.31 mM, and the Km for catalyzing TMB was 0.49 mM, demonstrating the high POD enzyme activity of Cu SAzymes. On this basis, Cu SAzymes were coupled with GOx to construct a cascade catalytic colorimetric method for sweat glucose detection. The detection limit of this method was as low as 0.06 mM, offering advantages of high sensitivity and good selectivity. To achieve intelligent detection, an LSTM neural network deep learning approach was introduced, where the relationship between the color images of the reaction system and glucose concentration was trained to achieve high-sensitivity detection of glucose in sweat.