A Novel Approach for Diagnosing Chronic Kidney Disease Using a Non-Invasive ECG Mechanism and Deep Learning Technique

Main Article Content

K. Chapa, B. Ravi

Abstract

Chronic kidney disease is one among extreme causes of increasing mortality rate. Usually, CKD diagnosis and the level of CKD identification is based on traditional serum creatinine concentration or GFR levels. Although they are reliable researchers are finding out Non-invasive methods so that the diagnostic process may become simple. ECG based CKD diagnosis is one among the non-invasive techniques upon which reasonable research is going on. Obviously, ECG is used in screening cardiovascular diseases and at the same time using the same ECG if we are able to identify the CKD it shall be more prolific to the public. In this paper, a novel mechanism is proposed in identifying the CKD level using ECG. A standard and well processed ECG dataset containing data of 10,646 subjects is taken for modelling the mechanism and computing a score which is used in turn in identifying the CKD level. A deep learning model has been used to concrete the process with which an accuracy of 99% has been obtained. Further to validate the proposed mechanism of score generation and CKD level identification, the process is applied on three datasets and obtained the prediction accuracy levels of 88.7%, 93% and 91.5% for the respective datasets which indicates the usage of ECG is quiet acceptable for CKD prediction.

Article Details

Section
Articles