Bimetallic Silver-Copper Micro Flower Nanozyme-Based Colorimetric Sensor Assisted by Artificial Neural Network for Point-of-Care Uric Acid Detection.

The current investigation aims to offer a ready-to-use biosensor with the aid of an artificial neural network (ANN)-powered point-of-care solution model for the detection of uric acid (UA). This mechanism is based on the enzyme-like activity of a bimetallic silver-copper microflower (Ag-Cu-MF) heterostructure that catalyzes UA oxidase to yield superoxide (*O2-) and hydroxyl (*OH+) radicals through the color change of 3,3',5,5'-tetramethylbenzidine. The synthesized Ag-Cu-MF exhibited strong catalytic performance with the respective Km (Michaelis-Menten constant) and Vmax (maximum reaction velocity) values of 0.3476 mM and 52.36 ×10-8 M s-1 of UA. Both visual observation and UV-Vis spectroscopy verified the colorimetric shift from greenish-blue to light blue. The ImageJ results provided a linear range for the UA concentrations of 0-600 μM, with a calculated limit of detection of 0.016, 0.012, 0.010, and 0.010 μM, respectively, in DI water, pH 4, pH 11, and SBF. Similarly, ANN modeling employed in the prediction of UA concentration revealed better performance by the combined red, green, blue (RGB) color-difference feature, with the optimized multilayer perceptron (MLP)-Wide-Reg architecture achieving a cross-validation R2 of 0.837 ± 0.023. On the independent hold-out test set, the ANN ensemble demonstrated a promising internal calibration performance (R2 = 0.874, RMSE = 70.9 μM, and MAE = 54.5 μM) that outperformed conventional linear regression (R2 = 0.722, RMSE ≈105 μM) baseline model on the same internal hold-out observations. These results highlight the potential of ANN-driven image analysis for quantitative UA prediction across the four tested matrices under controlled image-acquisition conditions. The overall outcome ensures the effectiveness of nanozymes and ANN in biomedical regression, especially when only limited data are available for acquiring a robust capture of nonlinear patterns.