Significance of the Science
Rapid and accurate diagnosis is critical for conditions like heart attacks, yet current lab tests require centralized facilities, trained staff, and costly equipment—barriers that limit access in low-resource or emergency settings. Paper-based sensors offer a promising solution due to their low cost and ease of use, but traditional versions struggle with sensitivity and the ability to measure multiple biomarkers at once. This study addresses those challenges by combining paper-based fluorescence assays with artificial intelligence to deliver lab-quality results quickly, affordably, and outside of traditional hospital labs.
Highlights
Multiplexed paper-based fluorescence vertical flow assay (fxVFA) developed for point-of-care testing
Simultaneous detection of three cardiac injury biomarkers: myoglobin, creatine kinase-MB (CK-MB), and heart-type fatty acid binding protein (FABP)
Deep learning algorithms trained to interpret fluorescent signals and quantify biomarker levels with high accuracy
Rapid results in under 15 minutes using only 50 μL of serum sample
High sensitivity detection limits below 0.52 ng/mL for all three biomarkers with minimal cross-reactivity
>0.90 correlation with standard ELISA tests and <15% variation across clinical samples
Summary
The study introduces a fluorescence-based vertical flow assay (fxVFA) that integrates paper microfluidics with deep learning for multiplexed biomarker detection. Using conjugated polymer nanoparticles as fluorescent labels and a mobile phone-based handheld reader, the platform simultaneously quantified myoglobin, CK-MB, and FABP in human serum samples. The system achieved detection limits of 0.52, 0.30, and 0.49 ng/mL, respectively, with accuracy validated across 46 independently tested cartridges. Neural network models optimized through feature selection enhanced quantification robustness, outperforming conventional methods. Results correlated strongly with ELISA measurements (R² values of 0.92–0.95) and maintained coefficients of variation below 15%. The fxVFA’s affordability, speed (<15 min), and handheld format make it a competitive point-of-care tool for diagnosing cardiac events and a broadly applicable framework for decentralized diagnostics.
Read more (open access): https://onlinelibrary.wiley.com/doi/10.1002/smll.202300617
