Classifying the status of CKD using Randomized Weighted Optimization Model

Abstract

chronic kidney disease (CKD) is a substantial health-care burden owing to its growing occurrence, enhanced danger of development to end-stage renal disease, and reduced mortality and morbidity prognosis. It is quickly accelerating to a worldwide well-being catastrophe. The main causes of CKD include high blood pressure and diabetes. The major function of the kidney is filtering waste from a human body. When kidneys fail, waste builds up in bodies and eventually leads to death. Researchers all around the globe utilize the Glomerular Filtration Rate (GFR) and kidney damage indicators to define CKD as a disorder that causes decreasing renal function over a period of time. A person having CKD is more likely to die early. Doctors have a tough time recognizing the several disorders associated with CKD early enough to avoid the condition. Data mining approaches have recently been exposed to considerable research in CKD diagnosis, with a focus on accuracy, either through the simplicity of illness by conducting feature selection in addition to pre-processing or not before classification. This research work determines creatinine level is a kind of blood metabolite that has a significant relationship with GFR and Blood Urea Nitrogen (BUN) BUN Creatinine Ratio (BCR). Since measuring GFR and BCR is challenging, this work focused in analysing creatinine value is used to determine Estimated GFR (EGFR) and BCR from serum creatinine indirectly utilizing the attributes available in the dataset whereas the 28 attributes are considered inclusive of “Gender” with 523 records. The EGFR and BCR computing results are used in this prediction analysis for providing an accurately categorizing the status of CKD which has been utilized by Randomized Weighted Optimization Model (RWOM). Furthermore, the proposed RWOM with Neural Network (NN) is compared to the RWOM with Logistic Regression (LR), NN, LR in terms of analysing the better categorization status of CKD from patient records.