K-Nearest Neighbor (K-NN) Method for Optimizing Data Training on Diabetes Diagnosis and Chronic

Abstract

Information technology has entered various fields, one of which is the health sector. Many researchers develop expert systems, medical records, and hospital registration systems. The diagnostic system is a concern for researchers because, with a diagnosis system, patients can consult through the system without visiting a doctor (expert). To make the diagnosis system,need medical record data from patients who have had previous treatment.The data will be used as a source of diagnostic of system knowledge as data training. Chronic diabetes patient data to be tested are called training data. The K-Nearest Neighbor (K-NN) method is used to detect diabetes and chronic complications. By developing this system, it is expected that the number of patients who have chronic complications and diseases that accompany Diabetes can be suppressed. The results of the K-NN method are very optimal if tested inpatient data training with optimal values of y = 0.73 and x = 0.62