The following is a summary of “Prediction of metabolic syndrome following a first pregnancy,” published in the December 2024 issue of Obstetrics and Gynecology by Kawakita et al.  

The prevalence of metabolic syndrome is growing in the United States, and prediction models using pregnancy data could help forecast future development.  

Researchers conducted a prospective study to develop machine learning (ML) models that predict the future development of metabolic syndrome using pregnancy data.  

They analyzed data from the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-Be Heart Health Study (nuMoM2b-HHS) (2010 to 2020), 127 variables from pregnancy were included, and the dataset was divided into test (30%) and training (70%) sets. Random forest and least absolute shrinkage and selection operator (LASSO) regression models were developed, compared by area under the receiver operating characteristic curve (AUC), and the final model was refined using SHapley Additive exPlanations (SHAP) values.  

The results showed 754 (17.8%) participants developed metabolic syndrome, with a higher AUC for the random forest model (0.878, 95% CI, 0.846–0.909) compared to the LASSO model (0.850, 95% CI, 0.811–0.888; P<.001). The final model, using the top 3 variables (high-density lipoprotein, insulin, and high-sensitivity C-reactive protein), achieved an AUC of 0.867 (95% CI, 0.839–0.895), which was not inferior to the actual model (P=.08), and an AUC of 0.847 (95% CI, 0.821–0.873) in the test set.  

They concluded that ML models developed during pregnancy can predict the risk of developing metabolic syndrome 2 to 7 years after delivery. 

Source: ajog.org/article/S0002-9378(24)00460-5/fulltext