Abstract
Persistent infection with high-risk human papillomavirus (hrHPV) is a major cause of cervical cancer. The effectiveness of current HPV-DNA testing, which is crucial for early detection, is limited in several aspects, including low sensitivity, accuracy issues, and the inability to perform comprehensive hrHPV typing. To address these limitations, we introduce MTIOT (Multiple subTypes In One Time), a novel detection method that utilizes machine learning with a new multichannel integration scheme to enhance HPV-DNA analysis. This approach may enable more accurate and rapid identification of multiple hrHPV types within a single sample. Compared to traditional methods, MTIOT has the potential to overcome their core limitations and offer a more efficient and cost-effective solution for cervical cancer screening. When tested on both simulated samples (to mimic real-world complexities) and clinical samples, MTIOT achieved F1 scores (the harmonic mean of sensitivity and specificity) of 98 % and 92 % respectively for identifying subtypes with a sample size ≥ 50, suggesting that it may significantly improve the precision of cervical cancer screening programs. This work with MTIOT represents a significant step forward in the molecular diagnosis of hrHPV and may suggest a promising avenue for enhancing early detection strategies and potentially reducing the incidence of cervical cancer. This study also underscores the importance of methodological innovation in tackling public health challenges and sets the stage for future clinical trials to validate MTIOT's efficacy in practice.