Machine Learning based Detection Technique to Predict the Survival of Patients with Chronic Kidney Diseases

Abstract

There are currently many people all around the world who are suffering from chronic kidney infections. Today, everyone is attempting to be health-conscious, even though, owing to overwork and a hectic schedule, one only pays attention to one's health when symptoms appear. A few factors, for example, dietary habits, temperature, and expectations for daily luxuries, cause large numbers of people to be afflicted unexpectedly and without knowledge of their condition. Finding persistent kidney disease is often intrusive, costly, time-consuming, and dangerous. The main reasons why many people die without receiving care are especially in many developing countries since resources are few. As a result, early diagnosis and recognition of illness remains important, particularly in non-industrialized countries where illnesses are typically studied in late stages. However, if it does not show any symptoms at all, or if it does not show any disease-specific symptoms, it is very difficult to find &predict the disease type, detect and prevent such a disease, and this could lead to permanent health damage as well as the formation of new diseases, but machine learning can be a hope, as it is the best way for prediction and disease analysis. We will utilize data from CKD patients with 14 variables, as well as several machine learning approaches such as Decision Tree, SVM, and CNN model. To create an efficient machine learning model with the highest accuracy (by comparing several machine learning models) in predicting whether or not a person has CKD and, if so, how severe it is.