Ebola deep wavelet extreme learning machine based chronic kidney disease prediction on the internet of medical things platform

Abstract

SummaryVarious significant methodologies have been developed in classifying chronic kidney disease (CKD). But still, there emerge certain drawbacks, including high storage space requirements, increased diagnosis time, high cost of computation, and degraded accuracy. Hence in the proposed research work, Ebola deep wavelet extreme learning machine (EDWELM) is proposed for the precise classification of CKD and non‐CKD. Initially, the accessed data are preprocessed by transforming categorical to numerical values and replacing missing values with median to remove unwanted distortions. In the feature selection process, the hybrid methodology of Darts game and Battle royale optimization called Darts Battle game optimizer is carried out to choose the most discriminative features for improving classification accuracy. The final step undertaken is the classification process, which is one of the important data mining applications for distinguishing the data classes. In the proposed EDWELM classification method, ELM based autoencoder, wavelet neural network, and Ebola optimization search algorithm are carried out for effective CKD classification. From the CKD dataset, the performance is analyzed to various terms like accuracy, F1 score, precision, recall, kappa, and balanced score. The accuracy of 99.83% is attained by the proposed EDWELM classification method, which is comparatively better than the existing approaches.