Feature Selection and Ensemble Entropy Attribute Weighted Deep Neural Network (EEAw-DNN) for Chronic Kidney Disease (CKD) Prediction

Initial prediction and appropriate medication are the ways to cure Chronic Kidney Disease (CKD) in the early stage of progression. The rate of accuracy in the classification algorithms focuses on the usage of exact algorithms used to select he features in order to minimize the dataset dimensions. The accuracy not only relies on the feature selection algorithms but also on the methods of classification, where it predicts the severities that are useful for the medical experts in the field of clinical diagnosis. To minimize the time for computation and to maximize the classifiers accuracy level, the proposed study, Ensemble Entropy Attribute Weighted Deep Neural Network (EEAw-DNN) classification was aided to predict Chronic Kidney Disease. The rate of accuracy of the EEAw-DNN is surveyed with the help of feature selection using data reduction. Hence Hybrid Filter Wrapper Embedded (HFWE) based Feature Selection (FS) is formulated to choose the optimal subset of features from CKD set of data. This HFWE-FS technique fuses algorithm with filter, wrapper and embedded algorithm. At last, EEAw-DNNbased algorithm used for prediction is used to diagnose CKD. The database used for the study is “CKD” which is implemented using MATLAB platform. The outputs prove that the EEAw-DNNclassifier combined with HFWE algorithm renders greater level of prediction when correlated to other few classification algorithms like Naïve Bayes (NB), Artificial Neural Network (ANN) and Support Vector Machine (SVM) in the prediction of severity of CKD. Datasets were taken from University of California Irvine (UCI) machine learning repository.