Machine learning-assisted dual-mode immunochromatography for high signal-to-noise ratio detection of aristolochic acid-I.

Developing high signal-to-noise ratio strategies for immunochromatography remains challenging yet highly desirable. Here, we developed a "off-on" dual-mode immunochromatographic assay (ICA) enabling both sensitive on-field colorimetric detection (CICA) and accurate fluorescence quantification (FQICA) of aristolochic acid-I. The assay utilized porous coordination network (PCN) as colorimetric signal, fluorescence quencher, and antibody carrier, paired with europium-based metal-organic frameworks (Eu-BTEC) as fluorescence donor. Benefiting from its porous architecture and tunable absorption wavelengths, PCN exhibited favorable antibody coupling ability, remarkable colorimetric intensity and near complete excitation and emission quenching of Eu-BTEC. The ICA showed limits of detection (LOD) at 0.45 ng/mL (CICA) and 0.08 ng/mL (FQICA), representing 12- and 65-fold improvements over the LOD of AuNPs-ICA. Furthermore, machine learning integration enabled precise contamination classification and quantification. This work not only established an efficient strategy for aristolochic acid-I rapid detection also paved the way for integration immunoassays with machine learning in food monitoring.