Postpartum hemorrhage (PPH) remains the leading cause of maternal mortality and a common obstetric complication. Rapid onset and severity can lead to hemorrhagic shock and fatal outcomes if not promptly managed. Major causes include uterine atony, placental abnormalities, birth canal trauma, and coagulation disorders. This study proposes a risk prediction model for PPH based on a Deep Belief Network (DBN), using relevant risk factors as input features. To address limitations from random initialization in the DBN, an improved Particle Swarm Optimization (IPSO) algorithm—featuring adaptive inertia weight and learning factors—was introduced to optimize network parameters. The optimized model, referred to as IDBN, was applied to predict PPH risk levels more accurately and efficiently. The results demonstrate the superior performance of the IDBN method, which achieved the highest accuracy and F1 score among all models evaluated. This approach offers a novel, data-driven method for early PPH risk identification and clinical intervention.
Kang et al. (Thu,) studied this question.