Deep learning-integrated dual-mode vertical flow assay for multiplexed cardiac biomarker detection
Abstract
We present a deep learning-enabled dual-mode, multiplexed vertical flow assay (dual-mode xVFA) that integrates colorimetric and chemiluminescence (CL) detection for multiplexed cardiac biomarker testing at the point of care. A compact, Raspberry Pi-based optical reader captures both colorimetric signals and CL emission from a single paper-based cartridge. This multimodal optical strategy enables wide-dynamic-range quantification of three biomarkers (cTnI, CK-MB, and NT-proBNP) from ~50 µL of serum within 23 min. Neural network-based analysis, trained and blindly tested using serum samples from 92 patients, demonstrated strong agreement (Pearson’s r > 0.96) with clinical reference assays in blinded testing. The system provides a cost-effective, portable, and scalable optical biosensor platform for high-sensitivity point-of-care diagnostics.