Deep learning-based chemiluminescent point-of-care assay for high-sensitivity troponin testing

We present a chemiluminescence vertical flow assay (CL-VFA) that achieves high-sensitivity detection of cardiac troponin I (cTnI), a gold standard biomarker for diagnosing myocardial infarction. The CL-VFA system integrates a chemiluminescent paper-based sensor, a portable and low-cost Raspberry Pi-based reader, and a deep learning-powered concentration inference algorithm. This system can rapidly measure cTnI from a 50 µL serum sample within 25 min, reaching a detection limit of 0.16 pg/mL and covering a wide dynamic range of six orders-of-magnitude in concentration. The CL-VFA exceeds the sensitivity of commercial benchtop analyzers by an order-of-magnitude while maintaining a compact and lightweight design. In a blinded clinical study using 66 patient serum samples, the CL-VFA showed a strong correlation with results from an FDA-cleared analyzer, highlighting its reliability and robustness. The platform's affordability, ease-of-use, and high analytical performance make it ideal for use in under-resourced settings, offering a transformative step toward democratizing high-sensitivity diagnostics.