Predicting hypertensive disorders in pregnancy using multiple methods: Models with the placental growth factor...

Area of Science:

• Perinatal medicine
• Biomarker research
• Predictive modeling in obstetrics

Background:

• Hypertensive disorders in pregnancy (HDP) pose significant risks.
• Placental growth factor (PlGF) is a recognized biomarker for HDP.
• Optimizing PlGF-based models is crucial for accurate prediction.

Purpose of the Study:

• To evaluate different variable selection and modeling techniques for HDP prediction using PlGF.
• To determine the optimal gestational age for PlGF measurement in HDP prediction.
• To identify the most effective model for incorporating PlGF in HDP risk assessment.

Main Methods:

• Compared step-logistic regression and Lasso for variable selection in PlGF models.
• Assessed model performance based on predictive accuracy and gestational age.
• Selected the optimal PlGF model and gestational age range for final prediction.

Main Results:

• PlGF models tested at 15-16 weeks showed improved prediction over non-screened values.
• Both step-logistic regression and Lasso achieved over 92% sensitivity.
• The Lasso method yielded a more effective final prediction model.

Conclusions:

• Variable selection methods significantly impact PlGF-based HDP prediction models.
• The Lasso method offers superior performance for HDP prediction using PlGF.
• An optimal HDP prediction model incorporating PlGF was identified for the second trimester (15-26 weeks).