Diagnosis and Prediction of Type-2 Chronic Kidney Disease Using Machine Learning Approaches

Abstract

The field of biosciences is to a large extent about electronic health records. These records provide consistent and sufficient information and knowledge generated from an enormous amount of data. Many methods of machine learning cooperate to detect kidney disease. Persistent kidney disease can be treated at an optimal time with renal replacement therapy or renal dialysis. This is a very crucial stage for any patient because there are very few chances for a successful operation. In this disease, the patient's kidneys are damaged, and fluid containment is less in the kidney due to smaller fluid intake or hereditary reasons. The blood cannot be filtered as needed, and it is not purified. Some patients have diabetes, blood pressure problems, or family history of diabetes type 2. The major reasons for heart disease are kidney failure, anemia, and excessively high potassium and calcium. CKD leads to chronic hyperglycemia and glycation. This results in many complications like heart disease, and micro and macro vascular diseases. Risk factors for this disease include hyperglycemia, hypertension, and smoking. In addition, the hemodialysis process affects the kidney. In very critical situations, there is a need for kidney transplantation to increase the chances of life. Early detection of CKD in patients provides them the opportunity for timely treatment that reduces the chances of progression of the disease. With the help of machine learning models, physicians achieve fast and consistent performance. This chapter proposes to diagnose and predict CKD through a classification analysis model by ML algorithms such as state vector machine (SVM), Random Forest, decision tree, and logistic regression. The dataset is taken from the University of California, Irvine (UCI), online source, which consists of 400 samples, out of which 26 attributes will be selected and extracted for better results. In this chapter, machine learning algorithms are used for early detection of CKD; some classifiers and classification metrics are applied to diagnose CKD. In this chapter, prediction of CKD at early stages is made, and we achieved greater accuracy by using different machine learning classifiers. It shows some classification techniques like data processing and other algorithms for the detection of CKD with the help of different classifiers like SVM, decision tree, logistic regression, and Random Forest. The goal of this research is to determine whether a limited dataset or a small extent of data is beneficial for predictive models. The results of this chapter focus on kidney patients in terms of prediction performance and are not suitable for clinical healthcare. They are just prediction by training and testing the dataset of CKD. The classifier SVM gives the best results in terms of classification accuracy, which is 97.46%.