基于Cox比例风险模型的子痫前期母儿结局分层及预后模型的建立OA

Objective:To investigate the independent risk factors for maternal and neo-natal adverse outcomes in preeclampsia,and to construct stratified prediction models for mater-nal and neonatal adverse outcomes based on the Cox proportional hazards model,so as to pro-vide references for the timing of clinical preeclampsia delivery.Methods:A retrospective study was conducted using clinical data from 280 patients with preeclampsia between 20 and 37 weeks of gestation at the First Affiliated Hospital of Guangxi Medical University.Prediction models for maternal and neonatal adverse outcomes were developed separately.Patients were categorized into groups with and without adverse outcomes.Independent risk factors were screened using LASSO(Least Absolute Shrinkage and Selection Operator)regression,and prediction models were constructed based on Cox proportional hazards regression.The predictive performance was evaluated by the area under the curve(AUC)of the receiver operating characteristic(ROC)curve.Calibration was assessed using calibration curves.Internal validation was performed using the bootstrap resampling method,and external validation was conducted on a dataset of 150 ca-ses from another hospital.Results:The maternal adverse outcome prediction model incorporated platelet count,creatinine,alanine aminotransferase,peak systolic blood pressure prior to admis-sion,24-hour urinary protein quantification,and stratified gestational age at admission.The mod-el's C-index was 0.701(95%CI:0.63～0.77).The areas under the curve(AUC)for predic-ting outcomes at 2,7,and 14 days of expectant management were 0.791,0.755,and 0.80,re-spectively.The neonatal adverse outcome prediction model incorporated stratified gestational age at delivery,gestational days at admission,fetal distress,mean arterial pressure at admission,and creatinine.This model achieved a C-index of 0.85(95%CI:0.82～0.87),with AUCs of 0.914,0.95,and 0.963 for predicting outcomes at 2,7,and 14 days of expectant management,respectively.Both internal and external validation indicated good predictive performance for the two models.Conclusion:Platelet count,creatinine,alanine aminotransferase,peak systolic blood pressure prior to admission,24-hour urinary protein quantification,and stratified gestational age at admission are independent risk factors for maternal adverse outcomes in preeclampsia.Strati-fied gestational age at delivery,gestational days at admission,fetal distress,mean arterial pres-sure at admission,and creatinine are risk factors for neonatal adverse outcomes.The constructed stratified prediction models for maternal and neonatal adverse outcomes in PE can effectively predict the risk of adverse outcomes,guide clinical decision-making regarding the timing of de-livery,and thereby reduce the risk of adverse outcomes to improve perinatal outcomes.