Identification and validation of a novel machine learning model for predicting severe pelvic endometriosis: A retrospective study

Abstract

This study aimed to explore potential risk factors for severe endometriosis and to develop a model to predict the risk of severe endometriosis. A total of 308 patients with endometriosis were analyzed. Least absolute shrinkage and selection operator (LASSO) was performed to identify the potential risk factors for severe endometriosis. Then, we used seven machine learning (ML) algorithms to construct the predictive models. Finally, SHapley Additive exPlanations (SHAP) interpretation was performed to evaluate the contributions of each factor to risk prediction. About 59.2% (183/308) of patients were diagnosed with severe endometriosis. The random forest (RF) model performed best in discriminative ability among the seven ML models, achieving an area under the curve (AUC) of 0.744. After reducing features according to feature importance rank, an explainable final RF model was established with six features. From the SHAP map, we found that the negative sliding sign had the greatest impact on the diagnostic performance of the RF model. This study provided a personalized risk assessment for the development of severe endometriosis, which may enable early identification of high-risk patients, facilitating timely intervention and optimized treatment strategies.

Keywords: Machine learning; Prediction model; SHAP; Severe endometriosis; Transvaginal ultrasound.

Fig. 1

Flowchart of the study design.

Fig. 2

Typical ultrasonography of related ultrasound data.

Fig. 3

Demographic and clinical feature selection using LASSO regression. (A) Feature Selection Process Using LASSO Regression. (B) Coefficient Profiles and L1 Norm Across Different Log(λ) Values.

Fig. 4

Performance of ML models to predict severe endometriosis. (A) Comparison of Accuracy and ROC-AUC for Machine Learning Models in Predicting Severe Pelvic Endometriosis. (B) ROC Curves of Seven Machine Learning Models for Predicting Severe Pelvic Endometriosis.

Fig. 5

(A) SHAP Values for Feature Importance in the final model. (B) Mean SHAP Values of Key Features Influencing the final model. (C) ROC Curves for the final model’s Performance on Training and Testing Sets.

Fig. 6

Local Shapley values for interpretation of the final model. (A) left OE; (B) right OE; (C) obliterated Pouch of Douglas; (D) waterfall plot of patients at high risk of severe endometriosis; (E) right OE; (F) Color Doppler Flow Imaging of right OE; (G) patients at low risk of severe endometriosis.