Fe-porphyrin MOF nanozyme-based colorimetric sensor assisted by machine learning for detecting antioxidants in fruit juices.

Antioxidants (AOs) are key bioactive components governing the nutritional quality and oxidative stability of fruit juices. However, their similar reducing properties make it challenging for conventional single-signal colorimetric methods to combine individual quantitative analysis with pattern recognition in complex food matrices. Herein, an Fe-porphyrin metal-organic framework (FeP-MOF) nanozyme was constructed using iron(III) meso-tetra(4-carboxyphenyl)porphyrin chloride (Fe-TCPPCl) as the functional ligand, and further integrated into a platform for high-throughput quantification and machine-learning-assisted pattern recognition. The porous MOF framework enhanced the accessibility and catalytic efficiency of Fe-porphyrin active sites. Based on AOs-mediated inhibition of the FeP-MOF/H2O2/TMB oxidation reaction, a microplate-based colorimetric sensor was developed for quantitative detection of ascorbic acid, glutathione, cysteine, gallic acid, and tannic acid, with limits of detection of 0.52-41.54 μM. Moreover, a multi-pH cross-reactive sensor array converted single absorbance signals into composition-sensitive multidimensional fingerprints. Coupled with linear discriminant analysis and hierarchical cluster analysis, the array achieved cross-validated classification accuracies of 100.0% for five antioxidants and 98.0% for five fruit-juice categories. Results also agreed well with high-performance liquid chromatography measurements. This work integrates nanozyme amplification, quantitative analysis of individual antioxidants, fingerprint sensing, and data-driven classification, providing a versatile strategy for high-throughput AOs analysis and compositional evaluation in complex food systems.