Imidazole derivative-regulated Cu-MOF nanozyme sensor array integrating machine learning for flavonoids detection and...

Flavonoids are important natural bioactive compounds with remarkable health benefits, requiring rapid and accurate detection for food nutrition assessment and quality control. However, it remains challenging to simultaneously discriminate multiple flavonoids in complex matrices. Taking inspiration from the tunable structures of MOF-based nanozymes and the ability of sensor arrays to simultaneously detect multiple substances, we herein propose an innovative four-channel colorimetric sensor array for flavonoid identification. Four copper-based MOF nanozymes (Cu-ImC₂N, Cu-ImC₂, Cu-ImCl₂, Cu-ImPh) were synthesized with different imidazole derivatives as ligands, showing significant differences in their peroxidase-like activities. Exploiting their specific catalytic activity variations at pH 6.0 and 6.5, the two most active nanozymes (Cu-ImC₂N and Cu-ImC₂) were strategically selected to construct the array. Upon processing the array data with linear discriminant analysis (LDA), the array achieved 100% accurate discrimination and quantitative detection of seven flavonoids, with limits of detection (LODs) ranging from 0.039 to 0.982 μM. Furthermore, comparisons of concentration-independent models built with multiple machine learning algorithms showed that the K-nearest neighbors (KNN) method outperformed LDA, effectively reducing interference from concentration variations. The sensor array was successfully applied to the analysis of commercial soy milk, reliably differentiating products from different brands and bean sources (e.g., soybean and mung bean). This strategy surpasses traditional single-analyte methods and holds substantial practical potential for the rapid on-site detection of complex food systems.