Objective:

To develop and internally validate a multivariable prediction model for persistent/recurrent HR-HPV positivity at 12 months using vaginal microecological and clinical characteristics.

Approach:

• Study Design: Retrospective study including women with baseline HR-HPV positivity and 12-month follow-up testing.
• Cohorts: 1,186 women divided into a training cohort (n = 830) and an internal validation cohort (n = 356).
• Predictor Selection: Candidate predictors selected using LASSO regression with ten-fold cross-validation.
• Model Evaluation: Model discrimination, calibration, and clinical utility evaluated using AUC, calibration curves, and decision curve analysis.

Key Findings:

• 29.8% of women had persistent/recurrent HR-HPV positivity at 12 months.
• Seven predictors retained: age, smoking status, HPV16/18 infection, vaginal pH, non-Lactobacillus-dominant microbiota, bacterial vaginosis, and moderate-to-severe local inflammation.
• HPV16/18 infection showed the strongest association (OR 8.564, 95% CI 5.383–13.955, P < 0.001).
• Final model AUC: 0.892 in training cohort, 0.821 in validation cohort.
• Calibration was acceptable in both cohorts.

Interpretation:

The developed model shows promising discrimination and acceptable calibration.

Limitations:

• Study conducted at a single center, limiting generalizability.
• External validation in independent cohorts is needed.

Conclusion:

The model may support individualized risk stratification; however, external validation in independent cohorts is needed.