The new era of risk assessment for hypertension in pregnancy: From clinical to biochemical markers in a comprehensive predictive model

Abstract

Objective

This study aims to develop and validate a model based on the weighted random forest (WRF) algorithm to predict early-onset preeclampsia (PE) and to assess the importance of various clinical and biochemical markers in early risk identification.

Materials and methods

This study was conducted at the Jiangxi Maternal and Child Health Hospital and involved 12,699 pregnant women from January 2019 to June 2022. Extensive clinical and biochemical markers were collected through prenatal care data, which were used to construct a predictive model for early-onset PE. The model was developed using the WRF and Logistic regression methods, and multivariable analysis was employed to identify markers significantly associated with the risk of PE.

Results

The relative importance of various markers was evaluated using the random forest (RF) model in a sample of 1200 patients diagnosed with PE. Blood pressure and pre-pregnancy body mass index (BMI) were identified as the most critical variables affecting the accuracy of the PE prediction model. The WRF model demonstrated higher predictive accuracy (AUC = 0.9614) than the Logistic regression model (AUC = 0.9138), highlighting its superiority in early risk identification for PE.

Conclusion

The WRF-based predictive model developed in this study effectively predicts the risk of early-onset PE, with blood pressure and BMI as vital predictive factors. These findings underscore the importance of employing a comprehensive predictive model for risk assessment in early pregnancy, facilitating early intervention and improving health outcomes for pregnant women and their newborns.