Development and validation of a predictive risk model for placental abruption in women with pre-eclampsia using SMOTE.

Abstract

Placental abruption is a serious complication in pregnant women with preeclampsia. This study investigated the risk factors for placental abruption and developed a predictive model using the Synthetic Minority Over-sampling Technique (SMOTE). A total of 280 pregnant women with preeclampsia treated between September 2020 and September 2025 were enrolled and classified into placental abruption and non-placental abruption groups. Logistic regression showed that parity ≥3, severe preeclampsia, anemia, polyhydramnios, and short umbilical cord were independent risk factors for placental abruption. ROC analysis demonstrated that the SMOTE-based model had better predictive performance than the original model, with a higher AUC (0.863, 95% CI: 0.817-0.901 vs 0.792, 95% CI: 0.739-0.838; z=2.163, P=0.031). The SMOTE-based model also showed higher F-score and positive predictive value, although its true positive rate was lower than that of the original model. These findings indicate that a SMOTE-based predictive model may provide useful support for early risk assessment and prevention of placental abruption in women with preeclampsia.