An improved various deep learning models to predict the kidney diseases in health data

Abstract

Smarter big data handling in Health care sector become indispensable, newlineas the need for identifying the risks of any chronic condition will save the life or newlinehelp in avoiding fatalities. As this quantum of data increases, the architectures and newlinetechnologies needed for handling those data might require continuous newlineimprovements. newlineFurthermore, the chronic level of kidney disease needs to have much newlineattention, since it could result in increased chances of casualities if left untreated. newlineHaving identified the challenges in predicting the chronic kidney disease such as: newlineDelayed consultation with nephrologists; Identification of late-stage chronic newlinekidney disease; Episodic out-patient follow up; Insufficient engagement of newlinepatients; and improper communication taking place between doctor and patient. newlineThe proposed work will have following three models to contribute in predicting newlinethe non-chronic as well as chronic level of kidney disease. newlineThese three models that are proposed in this research would be newlineconsidering all the precautionary diagnosing methodologies deployed for newlinepredicting the chronic and non-chronic kidney conditions by knowing the newlineevolution taking place from time to time ranging by the adoption of deep newlinelearning-based methodologies. newlineThe first phase of the work is a novel MR (Map Reduce) and pruning newlinelayer-based classification model using the Deep Belief Network (DBN). newlineThe Chronic Kidney Dataset (CKD) obtained from the UCI repository of learning newlinemachine is used. newlineThe second phase of work is a novel feature selection dependent newlineprediction model for diagnosing and acting on the chronic level of kidney diseases newlineto avoid sudden failure of it. newline