Machine learning-assisted early evaluation of aquatic products freshness based on UV-enhanced multi-enzymatic cascade

Early and rapid detection of the freshness of aquatic products is crucial for their quality control. Hypoxanthine is considered as a key metabolite during the early stage of spoilage in aquatic products. In this study, we investigated a machine learning-assisted colorimetric method for hypoxanthine detection, utilizing an ultraviolet (UV)-enhanced xanthine oxidase (XOD)/mimetic peroxidase cascade reaction. UV irradiation effectively enhanced the peroxidase-like activity of bimetallic Fe/Ni metal organic framework (Fe 7 Ni 3 MOF), thereby enhancing the efficiency of hypoxanthine detection. Results revealed that under UV irradiation conditions, the UV-enhanced XOD/Fe 7 Ni 3 MOF cascade reaction can detect hypoxanthine within 4–70 μmol/L, with a detection limit of 1.63 μmol/L. This method allowed for the detection of hypoxanthine in large yellow croakers and shrimps, achieving spiked recovery rates ranging from 92.16 % to 127.31 %. By capturing colorimetric color images using a smartphone and extracting nine chromatic information parameters, support vector machine regression (SVMR) model was used to predict hypoxanthine content, exhibiting a coefficient of determination for prediction (R 2 P ) of 0.946 and a ratio of performance to deviation of 3.79. This study offers a novel alternative method for the rapid detection of quality changes in aquatic products.