Chronic kidney disease prediction using different machine learning models

Abstract

A kidney's major purpose is to eliminate waste materials and excess fluids from the body through urine, which helps to maintain a stable chemical equilibrium in the body. Chronic kidney disease (CKD) is a serious global concern that is defined by a steady decline of kidney function over time. CKD affects over 14% of the world's population and is difficult to identify in its early stages. This disease is usually detected at the final or most critical stage in the human body, posing a significant risk to the human body and often resulting in the person's death. If the condition is identified early on, the patient's kidney function may be saved, allowing him or her to live a longer life. Machine learning has progressed to the point that we can now examine the medical records of individuals and detect chronic kidney disease in its early stages. On the CKD dataset from the UCI machine learning repository, this research examines the occurrence of CKD by creating ML models with 6 distinct classification algorithms. Before we can use machine learning techniques on the raw dataset, we must first process it and remove any duplicated or null variables before sending it to the models. After running the data through all of the models, it was observed that Random Forest and Extra Trees Classifier proved the highest accuracy of 98.33. The literature survey conducted before execution offered valuable insights and helped to shorten the execution time because we only chose algorithms with good accuracy.