Performance Analysis of Chronic Kidney Disease Detection Based on K-Nearest Neighbors Data Mining

Kidney diseases are a leading cause of death in the United States. According to the Centers for Disease Control and Prevention (CDC), in 2021, approximately 37 million US adults, or 1 in 7, are estimated to have chronic kidney disease (CKD), and most are undiagnosed. Moreover, Medicare costs for people with CKD were $87.2 billion in 2019. Thus, data mining has been used in the healthcare industry to assist authorities in providing patients with health information as well as identifying patients earlier. In this paper, data mining is implemented for the classification of laboratory data from CKD patients. The K-Nearest Neighbors (KNN) algorithm is proposed to train the machine learning model to detect CKD based on blood test lab results such as sugar count, white blood cell count, red blood cell count, hemoglobin, albumin, etc. The model also includes general factors such as age and blood pressure. From the obtained results, other machine learning methods produce inferior accuracy, such as linear regression and decision tree. By training the model on a dataset containing 400 different anonymous patients using KNN, the accuracy reaches 99%. Based on the prediction, around 40% of the patients are fully healthy. This paper aims to detect whether the patient has CKD or not, depending on lab results and general information about the patient