DESIGN OF A NOVEL FUZZY INFERENCE CONVOLUTIONAL NEURAL NETWORK (FICNN) BASED MATERNAL AND EMBRYO RISK ASSESSMENT SYSTEM FOR FECG AND USG SIGNALS

Abstract

Objectives: To design an intelligent system to identify the eight different risk factors to determine the state of the maternal and embryo health. Methods: A hybrid approach namely Improved Genetic Ant Lion Optimization (IGALO) which combines Genetic selection operator such as tournament and Ant Lion Optimization is proposed for optimal feature selection in the process of Maternal and embryo risk factor classification. A novel strategy has been proposed to explore the potential benefits of combining Deep Learning (DL) with fuzzy-based fusion techniques. Specifically, introduce fuzzy layers to the DL architecture. Finally, on the basis of training details, a proposed FICNN classifier identifies a class of data. Findings: For the purposes of experimentation, the simulation of the proposed FICNN classifier utilizes fine pre-trained datasets. Based on metrics, such as Accuracy, TPR, TNR, PPV, NPV, HM, Kappa, AUC, PRAUC, LogLoss, Detection Rate, Prevalence, Detection Prevalence and Balanced Accuracy, the results of the proposed FICNN classifier are assessed. The results obtained affirm the utility of FICNN to classify records within maternal and embryo surveillance systems that are computer-aided. Novelty: Our results suggest that the system is able to predict risk factors during the maternal delivery and then it gives the best accuracy by the proposed technique as Fuzzy Inference Convolutional Neural Network (FICNN) is very useful to reduce the maternal and embryo morality. The proposed technique achieved better classification performance classified the risk factors with Accuracy, TPR and TNR of 98.89, 100 and 98.77 for D1 also same equivalent Accuracy, TPR and TNR achieved by the other seven risk factors such as Caesarean, Hypoxia, Indication of childbirth, Diabetes, Preeclampsia, Hypertension, Meconium, respectively being highly adaptable to different laboratory settings, and easy integration into clinical practice.