A Novel Model for Diagnosing High-Risk Pregnancies Using Bayesian Belief Network Algorithm and Particle Optimization

Abstract

Introduction: Diagnosis of high-risk maternal pregnancy is one of the most important issues during pregnancy and can be of great help to pregnant mothers. Also, early diagnosis can reduce mortality and morbidity in mothers. Material and Methods: In this study, the data of 1014 pregnant mothers were used, which includes 272 people with high-risk pregnancies, 742 people with medium-risk and low-risk pregnancies. Also, the data include six independent variables. A combination of Bayesian belief network algorithms and particle optimization was used to predict pregnancy risk. Results: For validation, the data model was divided into two sets of training and testing based on the method of 30-70. Then the proposed model was designed by training data. Then the model for training and testing data was evaluated in terms of accuracy parameters 99.18 and 98.32% accuracy were obtained, respectively. It has also performed between 0.5 and 8% better than similar work in the past. Conclusion: In this study, a new model for designing Bayesian belief network was presented and it was found that this model can be useful for predicting maternal pregnancy risk.