Related Experiment Video
Updated: May 10, 2026
07:13
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
Multi-center-validated machine learning model for cervical cancer based on human papillomavirus genotyping results
Ziren Chen1, Yihui Zeng1, Lingxiao Luo1
1Obstetrics and Gynecology Hospital of Fudan University, Department of Cervical Disease Center, Shanghai, China.
Summary
This study developed an AI model using HPV genotyping to accurately predict high-grade cervical intra-epithelial neoplasia (CIN2+). The model improves cervical cancer screening by stratifying risk and reducing unnecessary colposcopy referrals.
Area of Science:
- Gynecology
- Oncology
- Medical Informatics
Background:
- Cervical cancer is a major cause of mortality, with human papillomavirus (HPV) testing crucial for screening.
- Current HPV testing guidelines broadly categorize high-risk types, potentially overlooking varying oncogenic risks.
- Precise risk stratification is needed for effective cervical cancer screening and management.
Purpose of the Study:
- To develop and validate a machine learning model for predicting cervical intra-epithelial neoplasia grade 2 or worse (CIN2+).
- To utilize HPV genotyping, cytology, and patient age for improved prediction accuracy.
- To enhance clinical decision-making in cervical cancer screening.
Main Methods:
- Retrospective analysis of 61,022 clinical records from 52,063 patients (Oct 2017-June 2023).
- Comparison of six machine learning algorithms (CatBoost, XGBoost, etc.) using AUC, accuracy, sensitivity, and F1 score.
- Multi-center validation including internal and external cohorts (Shanghai, Xiamen, Jinan).
Main Results:
- The CatBoost model achieved an AUC of 0.917, 86% accuracy, and 78% sensitivity for CIN2+ in internal validation.
- External validation demonstrated robust performance with accuracies ranging from 78% to 89%.
- An AI-guided strategy reduced colposcopy referrals by 14.3% while maintaining high sensitivity (96.8%) and NPV (99.18%).
Conclusions:
- A multi-center-validated machine learning model accurately stratifies CIN2+ risk.
- The model accounts for HPV genotype-specific pathogenicity, improving clinical decision-making.
- Developed AI tool (www.cervixcare.cn) provides individualized risk scores for cervical cancer screening.
