Portable Electrochemiluminescence Microarray Sensor for Machine Learning-Assisted Quantitative Analysis of Zearalenone

Zearalenone (ZEN) is an estrogenic mycotoxin produced by Fusarium fungi and is widely present in cereal-based foodstuffs including wheat, rice, and maize. The exposure to ZEN can cause severe endocrine disruption and reproductive disorders. Therefore, we develop a portable electrode electrochemiluminescence (ECL) platform based on a micropillar electrode design for image-based quantitative analysis of ZEN. In this study, luminol-functionalized Zr-based metal-organic frameworks (Zr@Lu-MOFs) modified on micropillar electrodes exhibited stable ECL emission and enabled controlled signal generation, while a miniaturized electrochemical workstation supported portable operation. ECL images acquired by a smartphone are converted into red-green-blue (RGB) feature data sets, which are subsequently analyzed using a two-step machine learning workflow. K-nearest neighbor (KNN) classification is employed for signal pattern discrimination, followed by Gaussian process regression (GPR) for quantitative concentration prediction. This image-driven analytical strategy allows ZEN to be quantified over a broad linear range from 0.0001 to 100 ng/mL with a limit of detection (LOD) as low as 0.23 pg/mL. Benefiting from the microarray architecture and data-driven analysis, the platform exhibits high signal uniformity, good reproducibility, and reliable quantitative performance in real samples. This study demonstrates that integrating ECL microarrays with image-level data analysis provides an effective route toward scalable and portable analytical platforms for food safety monitoring.