Construction of a Nomogram Prediction Model for Individualized Prediction of Cervical Cancer Risk: A Retrospective Study.

OBJECTIVE

To construct and validate a nomogram prediction model for the risk of cervical cancer.

METHODS

From May 2021 to September 2023, 1234 female patients who underwent cervical cancer examinations were included in the training set and divided into a cervical cancer group (97 cases) and a non-cervical cancer group (1137 cases). In addition, 528 women who underwent cervical cancer screening between October 2023 and October 2024 were included as the validation set. Multivariate logistic regression analysis was used to identify independent risk factors. R software was used to establish a nomogram prediction model. Moreover, ROC curve analysis, the Hosmer-Lemeshow (H-L) test, calibration plots, and decision curve analysis were used to evaluate the predictive performance of the model.

RESULTS

There was a statistically significant difference in the positivity of human papillomavirus (HPV) between the training set and the validation set (P<0.05). Education level of junior high school or below, menarche age ≤14 years, first sexual intercourse age ≤ 20 years, number of sexual partners > 2, long-term use of hormonal contraceptives, number of deliveries > 2, HPV positivity, cervicitis, and systemic lupus erythematosus were independent risk factors for cervical cancer (P<0.05). The AUCs of ROC curve for internal and external validation were 0.867 and 0.850, respectively. H-L test showed χ2 =11.541 and 5.769, with P=0.173 and 0.673, respectively. The ideal curve and calibration curve had a high degree of overlap, and the nomogram model demonstrated good discrimination and calibration. In the clinical decision curve, the net benefit of the nomogram model was higher within the threshold probability range of 0.02-0.78.

CONCLUSION

The Nomogram model based on nine identified factors provides a reliable tool to assist in the early detection and risk assessment of cervical cancer, which may help identify high-risk populations and guide targeted screening strategies in clinical practice.