Home    中文  
 
  • Search
  • lucene Search
  • Citation
  • Fig/Tab
  • Adv Search
Just Accepted  |  Current Issue  |  Archive  |  Featured Articles  |  Most Read  |  Most Download  |  Most Cited

Chinese Journal of Clinical Laboratory Management(Electronic Edition) ›› 2025, Vol. 13 ›› Issue (01): 17-26. doi: 10.3877/cma.j.issn.2095-5820.2025.01.003

• Automation and Information System • Previous Articles     Next Articles

Construction of a clinical prediction model for adverse pregnancy outcomes in patients with preeclampsia based on machine learning methods

Yu Yan1,2, Jianxin Zhang1,2, Hongwei Li1,2, Yuan Wei2, Yanzi Ding1,2, Mengyu Fu1,2, Xuewei Zhang1,2, Meilin Kan2, Enwu Yuan1,2,()   

  1. 1. Department of Laboratory Medicine, the Third Affiliated Hospital of Zhengzhou University, Zhengzhou Henan 450052, China
    2. The Third Clinical Medical College of Zhengzhou University, Zhengzhou Henan 450052, China
  • Received:2024-10-08 Online:2025-02-28 Published:2025-05-09
  • Contact: Enwu Yuan

Abstract:

Objective

To analyze the risk factors for adverse pregnancy outcomes in patients with preeclampsia (PE) and to develop and validate a predictive model.

Methods

This retrospective study included data from patients diagnosed with preeclampsia (PE) who delivered at the Third Affiliated Hospital of Zhengzhou University-Henan Provincial Maternal and Child Health Hospital from January 2021 to March 2024. Patients were categorized into adverse and non-adverse outcome groups based on the occurrence of adverse pregnancy outcomes. After screening, univariate and multivariate logistic regression analyses were used to evaluate clinical data. Independent risk factors identified were then utilized to develop a nomogram and construct a predictive model via R software 4.2.1. The receiver operating characteristic curve was calculated, and the model's discrimination, calibration, and clinical utility were assessed through the area under the curve (AUC) and Hosmer-Lemeshow (H-L) test. Internal validation of the predictive model was performed using the bootstrap resampling method and ten-fold cross-validation.

Results

A total of 472 PE patients were included in this study to construct the model, comprising 428 patients with adverse pregnancy outcomes and 44 without. Multivariate logistic regression analysis identified independent risk factors for adverse pregnancy outcomes in PE patients (P<0.05), including gestational age at onset ≤34 weeks, peak diastolic blood pressure during pregnancy ≥110 mmHg, abnormal umbilical artery blood flow, positive urine protein analysis, serum uric acid >369 μmol/L, and lactate dehydrogenase >246 U/L. Based on these independent risk factors, a clinical predictive model was constructed, yielding an AUC of 0.942 (95% CI:0.909~0.975). The optimal cutoff value was 0.836, with a specificity of 84.1% and sensitivity of 92.1%. The H-L goodness-of-fit test indicated good calibration of the model (χ2=4.969, P=0.761). Decision curve analysis shows that the model can achieve higher net benefits in clinical decision-making. The model validation shows that the model has good generalization ability and practical application value.

Conclusion

Based on clinical data, this study identified 6 independent risk factors to construct a predictive model for adverse pregnancy outcomes in preeclampsia patients. The model demonstrated strong predictive performance, with an AUC of 0.942 and good calibration (P=0.761), indicating substantial potential for clinical application.

Key words: preeclampsia, clinical data, pregnancy outcomes, risk assessment, risk factors, predictive model

京ICP 备07035254号-20
Copyright © Chinese Journal of Clinical Laboratory Management(Electronic Edition), All Rights Reserved.
Tel: 020-81341720 Fax: 020-37103505 E-mail: clinlab@cma.org.cn
Powered by Beijing Magtech Co. Ltd