Construction and Validation of a Model for Predicting Cervical Intraepithelial Neoplasia Grade II+: A Cross-Sectional...

Cervical cancer, as the leading malignant tumor among women globally, underscores the critical need for early screening; however, effective models for predicting cervical lesions remain lacking.

To construct a predictive model for cervical intraepithelial neoplasia II+(CINII+), and to compare the predictive performance of machine learning models integrating thinprep cytologic test (TCT) + human papillomavirus (HPV) testing with clinical data versus TCT combined with traditional clinical data for CIN II+.

Clinical data from women undergoing cervical cancer screening at Linquan Maternity and Child Healthcare Hospital (2020-2024) were collected, including TCT results, HPV status, cervical pathology, age, sexual history and other clinical data. Ten machine learning algorithms were applied to develop two predictive models: Model 1(TCT+HPV+clinical data) and Model 2(TCT+traditional clinical data). Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves and decision curve analysis (DCA).

Multivariate logistic regression analysis showed that HPV positivity, TCT indicates High-Grade Squamous Intraepithelial Lesion(HSIL), colposcopy result indicates a high-grade lesion and the first age of pregnancy as predictors of CINII+. Model 1 (TCT+HPV+clinical data) demonstrated significantly higher predictive efficacy than Model 2(TCT+clinical data), the difference in AUC is statistically significant. (P=0.006 in training set; P=0.035 in testing set).

The TCT+HPV-integrated model outperformed the TCT-only model in predicting CIN II+, supporting the incorporation of HPV testing into routine screening to enhance early diagnostic accuracy.