Machine learning-enabled smartphone CRISPR-Cas12a lateral flow platform for sensitive detection of circulating HPV...

Area of Science:

• Molecular Diagnostics
• Biotechnology
• Oncology

Background:

• Persistent high-risk human papillomavirus (HPV) infection is a primary cause of cervical cancer.
• Current PCR-based HPV detection methods require centralized labs, limiting point-of-care (POC) accessibility.
• CRISPR-Cas diagnostics with lateral flow assays (LFA) offer rapid detection, but visual interpretation can be subjective.

Purpose of the Study:

• To develop a smartphone-based CRISPR-Cas12a LFA platform for quantitative detection of circulating HPV DNA.
• To integrate an interpretable machine learning (ML) framework for objective and reliable HPV DNA analysis.
• To assess the platform's performance for point-of-care (POC) applications.

Main Methods:

• Developed a smartphone-based CRISPR-Cas12a LFA system with a standardized light-controlled enclosure for image acquisition.
• Utilized radiomics-inspired features and a multivariable logistic regression model for HPV DNA quantification.
• Trained the ML model on 150 plasma samples and validated it on an independent cohort of 60 samples.

Main Results:

• The optimized platform achieved 96.7% sensitivity and 100% specificity in detecting circulating HPV DNA.
• The system demonstrated superior performance compared to visual interpretation, especially for low-signal samples.
• Consistent and reliable operation was observed across various smartphone models, lighting conditions, and operators.

Conclusions:

• The integrated CRISPR-LFA platform provides accurate and reproducible detection of circulating HPV DNA.
• This technology shows significant potential for point-of-care (POC) cervical cancer screening.
• Further clinical validation is recommended for broader implementation in healthcare settings.