The Association of Serum Ferritin Levels with Endometriosis: A Machine Learning-Based analysis of NHANES Data, adjusted for Pet Exposure

Endometriosis has been linked to disturbances in iron metabolism; however, the association between circulating ferritin concentrations and prevalent endometriosis remains uncertain.We aimed to examine the association between serum ferritin concentrations and self-reported physician-diagnosed endometriosis and to evaluate machine-learning models for classifying endometriosis status.This cross-sectional study included 4,448 women aged ≥18 years from the 1999-2006 National Health and Nutrition Examination Survey, of whom 316 reported a physician diagnosis of endometriosis.Survey-weighted multivariable logistic regression and restricted cubic spline analyses were used to evaluate the association between serum ferritin concentrations and the prevalence odds of endometriosis.Following feature selection, decision tree, naïve Bayes, random forest, and support vector machine models were developed using a 70:30 training-test split and evaluated using 10-fold cross-validation.Model discrimination was assessed using the area under the receiver operating characteristic curve, and SHAP analysis was used to examine feature contributions.Serum ferritin concentrations were higher among participants with self-reported physician-diagnosed endometriosis than among those without endometriosis.After multivariable adjustment, participants in the highest ferritin quartile had significantly higher odds of reporting physician-diagnosed endometriosis than those in the lowest quartile (adjusted OR, 2.19; 95% CI, 1.28-3.74).Restricted cubic spline analysis suggested a nonlinear association between serum ferritin and endometriosis (P for nonlinearity = 0.001).Among the four machine-learning models, the random forest model showed the highest discrimination, with an AUROC of 0.862.Higher serum ferritin concentrations were positively associated with self-reported physician-diagnosed endometriosis.The random forest model showed promising internal discrimination; however, prospective studies and external validation are required before the model can be considered for clinical application.