Terbium (III) and adenosine monophosphate coordinated hydrogel for fluoroquinolones sensing by combining machine learning.

Nondestructive identification of fluoroquinolones distribution on food is an approach to simplify detection steps and improve detection efficiency. In this study, a fluorescent sensor based on terbium (III) and adenosine monophosphate coordinated hydrogel (TbCHy) has been synthesized, which is assisted by machine learning for intelligent and nondestructive detection of fluoroquinolones (FQs) in meat. Norfloxacin (NOR) was used as the model target, the TbCHy-based sensor performed a good linearity when NOR concentration ranged from 0.1 μM to 100 μM. It is selective toward eight FQs but there was no fluorescent response to sulfonamides and nitrofurans. The TbCHy-based sensor was pasted directly onto the spiked food matrices, combining with a trained convolutional neural network (CNN), NOR distributions were identifiable across the meat under 365 nm-UV light with good stability, specificity and accuracy. This approach eliminates the need for sample destruction or degradation and enables in situ monitoring of fluoroquinolones contamination distribution.