OP11. 07: Machine learning prediction model for gestational diabetes in twin pregnancies: a population based study

摘要: ObjectivesTo assess prevalence and risk factors of gestational diabetes (GDM) in twin compared with singleton pregnancies and to develop a prediction model to quantify the risk of developing GDM in twin pregnancies.MethodsPopulation-based study using CDC birth data from 2016-2021. Higher-order pregnancies and pre-pregnancy diabetes were excluded. Multivariable regression analyses were performed first to assess the association between GDM and candidate predictors and second to assess risk of GDM in twin compared with singletons, adjusted for the significant risk factors. All potential predictors were evaluated in two prediction models, logistic regression and machine learning (ML). The predictive accuracy was assessed by AUROC. Non-parametric test was used to compare AUROC of both models and given the large sample size, P value< 0.001 was considered statistically significant.ResultsTotal of …