JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Development and Validation of a Machine Learning-Based Early Warning Model for Lichenoid Vulvar Disease: Prediction Model Development Study

  • 0Department of Obstetrics and Gynecology, West China Second Hospital, Sichuan University, Chengdu, China.
Journal of Medical Internet Research +

|

November 22, 2024

|

November 22, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

This study identifies key risk factors for lichenoid vulvar disease (LVD) in women, developing a Gradient Boosting Machine (GBM) model for early detection. The GBM model effectively predicts LVD risk, aiding in timely intervention and prevention strategies.

Area Of Science

  • Gynecological Health
  • Dermatology
  • Predictive Modeling in Medicine

Background

  • Lichenoid vulvar disease (LVD) presents complex risk factors, necessitating personalized risk assessment models.
  • Existing research on LVD risk factors and predictive models is limited both nationally and internationally.
  • This study is the first systematic investigation into LVD risk factors and the development of a predictive model.

Purpose Of The Study

  • To identify and analyze risk factors associated with LVD in women.
  • To construct an evidence-based warning model for early LVD risk detection in high-risk populations.
  • To enhance clinical identification, prediction accuracy, and diagnostic proficiency for LVD.

Main Methods

  • Retrospective analysis of 2990 patients (1218 controls, 1772 LVD cases) from West China Second Hospital (2013-2017).
  • Collection and comparison of routine examination data to identify significant risk factors.
  • Development and evaluation of six predictive models (logistic regression, random forests, GBM, adaboost, XGBoost, CatBoost) using ROC curves and AUC.

Main Results

  • Significant risk factors for LVD include vaginitis, urinary incontinence, environmental humidity, spicy diet, coffee intake, sleep duration, diabetes, smoking, autoimmune diseases, menopause, and hypertension.
  • The Gradient Boosting Machine (GBM) model demonstrated superior predictive performance (AUC, accuracy, sensitivity, F1-score) compared to other models.
  • Menopausal status and autoimmune diseases were identified as the strongest positive predictors of LVD.

Conclusions

  • A predictive warning model for LVD, based on evidence-based medicine, can effectively identify high-risk individuals.
  • Early identification facilitates timely preventive measures, crucial for reducing LVD incidence in women.
  • The developed model holds significant potential for clinical application and reducing the overall disease burden.

Keywords: