Prospects of artificial intelligence in multimodal integration of methylation and imaging features for precision cervical cancer screening

Cervical cancer remains a major public health threat to women worldwide, and early screening is critical for reducing its incidence and mortality. Although human papillomavirus (HPV)-based screening strategies have achieved significant success, current methods still face limitations in specificity and risk stratification, leading to over-referral, missed high-grade lesions (HSIL+), and inefficient allocation of healthcare resources. Therefore, more objective and precise molecular markers and screening strategies are needed to optimize early detection of cervical cancer. DNA methylation abnormalities are early and relatively stable epigenetic events in cervical carcinogenesis, showing potential value in precise screening. DNA methylation markers, such as PAX1, ZNF582, and other candidate genes, can be used for risk stratification in HPV-positive populations. With the advancement of high-throughput sequencing, methylation information can be analyzed jointly with HPV status and pathological grading, and integrated with multi-omics data including genomics and transcriptomics, providing new perspectives for systematic characterization of cervical cancer at the molecular level. Meanwhile, the digitization of medical imaging, including cervical cytology, colposcopy, and magnetic resonance imaging, enables quantitative extraction and objective analysis of imaging features. Artificial intelligence (AI) provides a computational framework for integrating methylation-based multi-omics data with imaging features and identifying clinically relevant patterns in high-dimensional, heterogeneous datasets. This review summarizes methylation-based multi-omics strategies and recent advances in AI-driven multimodal integration for precision cervical cancer screening. Further data accumulation, methodological standardization, and clinical validation may support individualized screening and early intervention.