Deep Learning-Enhanced Hand-Driven Microfluidic Chip for Multiplexed Nucleic Acid Detection Based on RPA/CRISPR

Area of Science:

• Biotechnology
• Molecular Diagnostics
• Public Health

Background:

• Early detection of high-risk human papillomavirus (HR-HPV) is critical for cervical cancer prognosis.
• Current PCR-based methods lack accessibility, hindering screening in resource-limited areas and contributing to high mortality rates.

Purpose of the Study:

• To develop a portable, user-friendly system for multiplexed HR-HPV nucleic acid detection.
• To improve accessibility and efficiency of cervical cancer screening, particularly in underserved regions.

Main Methods:

• Integration of Recombinase polymerase amplification (RPA), CRISPR detection, hand-driven microfluidics, and an AI platform (R-CHIP).
• Utilized a smartphone microimaging system with the ResNet-18 deep learning model for automated result readout.
• Optimized for high sensitivity (10^-17 M for HPV-16, 10^-18 M for HPV-18) and accuracy.

Main Results:

• R-CHIP provides results in under an hour with simple manual operation.
• Achieved over 95% accuracy in 300 clinical sample tests for HPV-16 and HPV-18.
• Smartphone-based AI readout demonstrated initial prediction accuracies of 96.0% for HPV-16 and 98.0% for HPV-18.

Conclusions:

• R-CHIP is a sensitive, accurate, and accessible platform for community-level HR-HPV screening.
• The system holds significant potential for early cervical cancer detection and prevention in resource-constrained settings.
• The integrated approach offers a model for developing similar diagnostic tools for other diseases.