Breaking the Single-Mode Bottleneck of Natural Dyes via MOF Confinement for Intelligent Lateral Flow Immunoassays

Natural dyes offer biocompatibility and vivid color but are hindered in lateral flow immunoassays (LFIA) by weak signal, poor stability, and single-mode readout. Herein, we propose a Zr-MOF confinement strategy by encapsulating CA within spherical mesoporous UiO-66-NH2 via hydrogen bonding, π−π stacking, and electrostatic interactions to enable intelligent dual-mode LFIA. This porous scaffold, with an encapsulation efficiency of 66.91%, not only concentrates the chromophore and protects it from environmental stress, but also enhances antibody conjugation and creates strong spectral overlap with Arg/ATT-Au NCs, triggering an inner-filter effect (IFE) that enables highly efficient fluorescence quenching. The resulting dual-mode LFIA delivers LOD (limit of detection) of 0.399 ng mL−1 (colorimetric) and 0.298 ng mL−1 (fluorescent), which are approximately 9.95- and 13.33-fold more sensitive than conventional Au NPs-based assays, with recoveries of 85.5−110.3% in food matrices. To fully harness the complementary signals, a hybrid KNN-RF model processes the dual-channel data, achieving 100% classification accuracy and an R2 of 0.9975 for concentration prediction, thus eliminating operator subjectivity. This work demonstrates that MOF encapsulation transforms a fragile natural dye into a robust bifunctional probe, and when paired with machine learning, offers a reliable, intelligent pathway for point-of-care mycotoxin monitoring.