关键词: 高血压, 妊娠性, 妊娠结局, 收缩压, 随机森林预测模型

Abstract:

Objective To analyze the risk factors for adverse pregnancy outcomes in patients with hypertensive disorders of pregnancy (HDP), construct and validate a random forest prediction model by integrating routine clinical indicators, and evaluate the predictive value of multiple combined indicators. Methods A total of 152 patients with HDP who received diagnosis and treatment at the First People’s Hospital of Zhumadian from August 2022 to August 2025 were included. According to pregnancy outcomes, they were divided into an adverse outcome group (n=48) and a good outcome group (n=104). The incidence of adverse pregnancy outcomes and general clinical data were analyzed. Multivariate logistic regression was used to identify independent risk factors, and a random forest prediction model was constructed. The total sample was randomly split in a 7∶3 ratio into a training set (n=106) and a validation set (n=46). The training set was used to build the model, and the validation set was used to evaluate model performance. Results Among the 152 patients with HDP, 48 had adverse pregnancy outcomes, with an overall incidence of 31.58%. Compared with the good outcome group, the adverse outcome group had an earlier gestational age at onset and higher proportions of irregular antenatal care, systolic blood pressure (SBP), umbilical artery systolic/diastolic ratio (S/D), pentraxin-3 (PTX-3) level, and serum soluble fms-like tyrosine kinase-1 (sFlt-1)/placental growth factor (PLGF) ratio (P<0.05). Multivariate logistic regression confirmed that irregular antenatal care, SBP, umbilical artery S/D, elevated PTX-3 level, and increased sFlt-1/PLGF ratio were independent risk factors for adverse pregnancy outcomes in patients with HDP, whereas later gestational age at onset was a protective factor (P<0.05). Using these 6 independent risk factors from the multivariate logistic regression analysis as predictors, a random forest model was constructed. The results showed that the area under the curve(AUC) for predicting adverse pregnancy outcomes in patients with HDP was 0.874(95%CI: 0.829-0.918), with a sensitivity of 78.20%(95%CI: 0.726-0.851) and a specificity of 79.20%(95%CI: 0.754-0.896). In addition, based on the Gini coefficient importance scoring method, the contribution ranking from highest to lowest was sFlt-1/PLGF (51.676), gestational age at onset (42.945), umbilical artery S/D (40.135), SBP (34.866), PTX-3 (25.239), and irregular antenatal care (22.689). Conclusion Gestational age at onset, irregular antenatal care, SBP, umbilical artery S/D, PTX-3 level, and sFlt-1/PLGF are all influencing factors for adverse pregnancy outcomes in patients with HDP. The random forest prediction model established based on these variables demonstrated good discrimination and calibration, providing a potential tool for early identification of high-risk patients in clinical practice.

Key words: hypertension, pregnancy-induced, pregnancy outcome, systolic blood pressure, random forests prediction model