Objective To explore the risk factors for severe postpartum hemorrhage (SPPH) in patients with postpartum hemorrhage (PPH) after vaginal delivery and establish a predictive model. MethodsA retrospective analysis was conducted on 551 parturients with postpartum hemorrhage who underwent vaginal delivery at the Obstetrics and Gynecology Department of Shanxi Children's Hospital from January 2021 to December 2023. The study subjects were divided into mild postpartum hemorrhage group A (postpartum hemorrhage volume of 500-1 000 mL) and severe postpartum hemorrhage group B (postpartum hemorrhage volume ≥1 000 mL) based on the amount of postpartum hemorrhage. The causes and related factors of prenatal and postpartum hemorrhage in the two groups were compared, and univariate analysis and Lasso regression analysis were used. A Logistic multiple regression model was used to screen the influencing factors of postpartum hemorrhage, establish a line chart model for predicting postpartum hemorrhage risk, and evaluate the diagnostic efficacy of the prediction model using receiver operating characteristic (ROC) curve, calibration curve, decision curve analysis (DCA), and clinical impact curve (CIC).Results Univariate analysis showed that a history of previous intrauterine procedures, assisted reproduction (IVF), gestational hypertension, labor analgesia, oxytocin induced labor, perineal lateral resection, and placental residue had an impact on severe postpartum hemorrhage (P＜0.05). Further use of Lasso regression analysis was used to screen for Logistic multivariate regression analysis variables. The results showed that six risk factors were included in the multivariate Logistic regression analysis. Ultimately, a history of uterine cavity surgery, gestational hypertension, labor analgesia, oxytocin induced labor, perineal lateral resection, and placental retention were all independent risk factors for severe postpartum hemorrhage (P＜0.05). A column chart model for predicting the risk of severe postpartum hemorrhage was developed based on independent risk factors. The calibration curve of the model was close to the ideal curve, with an area under the ROC curve (AUC) of 0.741 (95% CI: 0.6871～0.7942). The clinical decision curve and clinical impact curve displayed thresholds between 0.05～0.5, indicating that the model had good net benefits and good application value. Conclusion The establishment of a column chart model for predicting the risk of severe postpartum hemorrhage during vaginal delivery can effectively predict the risk of such bleeding, and has certain clinical practical value