Boron and nitrogen co-doped carbon dots for fructose sensing: selective visual readout and machine learning-assisted...

Fructose levels vary widely across food products, creating a demand for simple and intuitive on-site detection. Here, boron and nitrogen co-doped carbon dots (BN,N-CDs) were prepared by a one-pot hydrothermal route from 3-aminophenylboronic acid and L-asparagine for fructose sensing. BN,N-CDs showed a more pronounced fluorescence response to fructose than to other tested sugars, enabling rapid visual screening at high levels. Under optimized conditions, the probe afforded quantitative fluorometric determination with a linear Stern-Volmer response over 0-800 μM and a detection limit of 3.10 μM. The method was validated in mixed-sugar systems with recoveries of 101.40-108.93% and successfully applied to commercial honey samples. In addition, multi-view fluorescence imaging combined with machine learning-assisted analysis enabled semi-quantitative readout. This work provides a practical strategy for visual fructose sensing in commercial honey samples.