Predictive Model for Pulmonary Embolism in Pregnant and Postpartum Women: A 10-Year Retrospective Study

Abstract

Background: Pulmonary embolism (PE) in pregnant and postpartum women is fatal, and risk assessment is crucial for effective and safe management, the aim of this retrospective study was to establish a nomogram for predicting the risk of PE in pregnant and postpartum women. Methods: Totally 343 subjects suspected of PE at the Obstetrics Department of Affiliated Dongyang Hospital of Wenzhou Medical University from January 2012 to December 2021 were retrospective analyzed in our study. Pregnant women suspected of PE and who underwent computed tomographic pulmonary angiography examination were included in the study. The least absolute shrinkage and selection operator regression technique was used to select the best prediction features, and multivariate logistic regression is used to build the prediction model. Bootstrap resampling 1000 times was used to validate the model visualized by nomogram. Evaluate the performance of the model from three aspects: identification, calibration and clinical utility. Results: Our predictive model indicated that chest tightness, anhelation, lactate, and D-dimer were associated with PE. The area under the receiver operating characteristic curve of the model was 0.836 (95% CI: [0.770-0.902]), indicating that our model had a good differential diagnostic performance. Good consistency between prediction and real observation was presented as the calibration curve. Decision curve analysis indicated that our model had a good net clinical benefit. Conclusions: We developed a novel numerical model for selecting risk factors for PE in pregnant and postpartum women. Our results may help obstetricians and gynaecologists to develop individualized treatment plans and PE prevention strategies.

A Practical Nomogram for Predicting the Bleeding Risk in Patients with a History of Myocardial Infarction Treating with Aspirin

“…Multiple imputation techniques were used for handling missing predictor values. 21 The data was divided into a training set (70% of the data) and a testing set (remaining portion) for model training and evaluation. The classification model was trained using the training set, and its performance was assessed using the testing set.…”