Smartphone-integrated portable microfluidic platform for liver biomarker quantification using deep learning.

Area of Science:

• Biomedical Engineering
• Medical Diagnostics
• Mobile Health Technology

Background:

• Accurate liver biomarker testing is vital for diagnosing and monitoring hepatic dysfunction, especially in underserved regions.
• Current methods often require centralized labs, limiting accessibility in resource-constrained settings.

Purpose of the Study:

• To develop a novel, smartphone-integrated colorimetric sensing platform for quantitative liver biomarker analysis.
• To enable decentralized and cost-effective liver function assessment using microfluidics, deep learning, and mobile health.

Main Methods:

• A stereolithography (SLA) 3D-printed microfluidic flow cell was designed for low reagent consumption and optical clarity.
• Chromogenic reactions for biomarkers were imaged using various smartphones within a controlled lighting enclosure.
• A convolutional neural network (CNN) analyzed the images for quantitative biomarker estimation, with a smartphone adaptability framework for cross-device consistency.

Main Results:

• The platform achieved clinically relevant detection ranges for direct/total bilirubin (0.1-20 mg/dL) and ALT/AST (10-300 U/L).
• Limits of detection were as low as 0.1 mg/dL for bilirubin and 2.5 U/L for AST.
• High accuracy was demonstrated with an average R² of 0.997 and repeatability with coefficients of variation under 3%.

Conclusions:

• This innovative, cost-effective, and portable system provides precise liver function assessment.
• The smartphone-integrated platform is ideal for rural healthcare and mobile diagnostics, enhancing early disease detection and monitoring.