Related Experiment Video
Updated: Sep 6, 2026
An Experimental Paradigm for the Prediction of Post-Operative Pain (PPOP)
Published on: January 27, 2010
Establishment of an antepartum predictive model for postpartum hemorrhage in preterm delivery: a retrospective
Qiuping Liao1,2,3,4, Baomei Xu5, Lin Lin1
1College of Clinical Medicine for Obstetrics & Gynecology and Pediatrics, Fujian Medical University Fujian Maternity and Child Health Hospital, Fuzhou, Fujian, China.
Background:
Postpartum hemorrhage (PPH) occupies a prominent position among severe maternal morbidity (SMM) in preterm pregnancies, yet current research on predicting PPH in preterm delivery remains limited.
Aim:
To establish an antepartum predictive model for postpartum hemorrhage in preterm delivery.
Methods:
This study included 728 singleton pregnant women. Clinical data of women were collected. Logistic regression analysis was performed to identify risk factors for PPH in preterm delivery, and a nomogram prediction model was established. Additionally, based on the basic principles and methods of the Back Propagation (BP) neural network and Garson algorithm, the importance ranking of variables was conducted.
Results:
Number of induced abortions≥2 [odds ratio (OR) = 2.067, 95%CI:1.192-3.585], gestational diabetes mellitus (GDM) (OR = 1.702, 95% CI: 1.111-2.607), placenta accreta (OR = 7.443, 95% CI: 3.955-14.005), placenta increta (OR = 60.170, 95% CI: 18.271-198.147), anemia (OR = 1.941, 95% CI: 1.303-2.891), and polyhydramnios (OR = 3.539, 95% CI: 1.202-10.417) were identified as risk factors for PPH in preterm delivery. The nomogram prediction model was constructed. The receiver operating characteristic (ROC) curve area under the curve (AUC) for the predictive model was 0.770. The Hosmer-Lemeshow test statistic χ2 = 1.708, P = 0.789. Following internal validation, the bias-corrected line approached the ideal line, with a consistency index (C-index) of 0.770. Using a BP neural network and Garson algorithm, the top three risk factors most significantly influencing PPH in preterm delivery were identified as placenta increta, placenta accreta, and polyhydramnios.
Conclusion:
This predictive model demonstrated moderate predictive ability in the derivation cohort, and its validation similarly showed acceptable performance. This model may be employed to identify women with high-risk factors of PPH in preterm delivery and to facilitate antepartum decision-making clinically.
