PURPOSE: Cervical cancer (CC), a common female malignancy, has been linked to alterations in DNA methylation. This study employed an integrated "dry-wet lab" strategy combining bioinformatics, machine learning, and experimental validation to identify novel methylation biomarkers for CC. METHODS: Methylome and transcriptome data from the TCGA and GEO cohorts (n=349 discovery, n=414 validation) were analyzed to identify differentially methylated CpGs. The top candidates were validated by pyrosequencing, methylation-specific PCR, and quantitative assays. Diagnostic models were developed, and functional studies were performed for the target markers. RESULTS: Eighteen differentially methylated CpGs were identified, with five top candidates (three in the ZSCAN18 promoter) showing diagnostic potential. ZSCAN18 promoter methylation levels and positivity rates were significantly greater in CC tissues than in normal tissues (p<
0.05), reaching 77.8% (21/27) in ThinPrep cytology test (TCT) samples. The ridge regression diagnostic model achieved an AUC of 0.9421 in the validation cohort. Similarly, ZSCAN18 overexpression suppressed CC cell proliferation (p<
0.05). CONCLUSIONS: This study established a rapid, effective and systematic systemic research strategy to screen novel methylation markers for CC. ZSCAN18 promoter methylation correlates with cervical lesion severity, and the diagnostic model enhances the diagnostic ability. These findings highlight the dual role of ZSCAN18 as a diagnostic marker and potential therapeutic target.