Classification of Imbalanced Fetal Health Data by PSO Based Ensemble Recursive Feature Elimination ANN

Electrocardiogram (CTG) is a simple and low-cost option to assess the health of the fetus. However, the number of normal fetuses is larger than the number of abnormal fetuses, leading to imbalances in CTG data. Existing studies have attempted to optimize the data processing or model training process by integrating machine learning methods with optimization algorithms. However, the effectiveness of features and appropriate selection of machine learning method creates new challenges. This study proposed an comprehensive method that considers the feature effectiveness and data imbalance issue. The proposed method uses the Particle Swarm Optimization (PSO) algorithm to optimize the parameters of the Edited Nearest Neighbours (ENN), Recursive Feature Elimination (RFE), and Artificial Neural Network (ANN) algorithms to find the optimal combination of the parameters of the three algorithms to further improve the accuracy of the fetal health prediction and reduce the cost of tuning. Experimental results show that the algorithm proposed in this paper can effectively solve the imbalance of CTG data, with a classification accuracy of 0.9942 and a kappa measure of 0.9783, which can effectively assist doctors in diagnosing fetal health and improve the quality of hospital visits.