Dual-mode detection of tetracyclines in animal-derived food samples based on MOF-on-MOF heterostructure fluorescent sensor coupled with machine learning algorithms.

Tetracyclines (TCs) that accumulate in food enter the human body through the food chain, leading to drug resistance and health risks. This study synthesized a heterostructure fluorescent sensor, Zn-TCPP-MOF/NH2-MIL-53(Al) (abbreviated ZTA) for TCs detection. A dual-mode detection system was established by integrating fluorescence spectroscopy with machine learning algorithms and fluorescence imaging with YOLOv5, enabling precise detection of TCs. The π-π stacking reduced ZTA electron density to promote electron transfer, inducing specific recognition of TCs with a low detection limit of 0.012 μmol/L. Machine learning algorithms of k-means clustering, hierarchical cluster analysis, support vector machine, random forest, logistic regression, and decision tree achieved 100.0%, and YOLOv5 reached 99.8% precision for TCs identification. The ZTA sensor achieved recovery rates between 95.40%-108.50% for TCs detection in animal-derived foods. This study facilitated the design of a MOF-on-MOF heterostructure sensor and provided an intelligent detection strategy for the rapid on-site monitoring of antibiotic residues.