Personalized risk stratification of preeclampsia in nulliparous women: retrospective data analysis and mathematical modeling

Prenatal prediction is a crucial stage in pregnancy management, which is aimed at personalized risk stratification of complications and improvement of perinatal outcomes. The use of multifactorial analysis and mathematical modeling enables to develop the effective algorithms for predicting preeclampsia in nulliparous women, contributing to a more differentiated approach to pregnancy management.The objective: to develop a mathematical prognostic model for personalized risk assessment of preeclampsia in nulliparous women by analyzing retrospective data and identifying the most significant anamnestic and biophysical predictors.Materials and methods. A retrospective cohort study was conducted at the clinical base of the Department of Obstetrics and Gynecology No. 3 of Bogomolets National Medical University – MNPE “Kyiv City Maternity Hospital No. 3” from 2018 to 2023. The analysis included data from 140 nulliparous women, among whom 88 (62.9%) patients were diagnosed with preeclampsia, while 52 (37.1%) women had no this complication.Results. A personalized analysis of the relationship between preeclampsia development and potential risk factors was performed. Thirteen factors which were associated with an increased risk of preeclampsia were identified, including age at menarche, excessive body weight, cardiovascular diseases, diabetes mellitus, prior COVID-19 infection, varicose veins, menstrual cycle disorders, inflammatory diseases of the pelvic organs, ovarian cysts, threatened pregnancy loss, placental hyperplasia, placental cysts, and high-resistance blood flow in the uterine arteries. Based on a multifactorial analysis using logistic regression, a prognostic model for preeclampsia risk was developed.Conclusions. Personalized risk factors for preeclampsia in nulliparous women have been identified. The developed model allows to predict preeclampsia risk with a sensitivity of 71.6% (95% confidence interval (CI) [61.0–80.7]) and a specificity of 78.8% (95% CI [65.3–88.9]).