Prediction and Severity Estimation of Diabetes using Datamining Techniques

Data mining plays an efficient role in prediction of diseases in health care industry. Diabetes is one of the major global health problems. Diabetes is a metabolic disease where the improper management of blood glucose levels led to risk of generating abnormalities in functioning of critical organs like heart attack, kidney, eye diseases etc. In this study, we have proposed a model based on data mining techniques for predicting diabetes mellitus. Based on a series of preprocessing procedures, the model is comprised of two parts, the K-means algorithm and the logistic regression algorithm. The Data set is collected from Pima Indian diabetes dataset containing various attributes like Age, Sex, BMI, and Test Results of diabetes. The proposed model that consisted of both cluster and class method ensured the enhancement of prediction accuracy. Early prediction and estimation of Severity may prevent developing high risk and may use to prevent death from severe diabetes. The logistic regression algorithm efficiently predicts the diabetic person and estimate the severity of the human organs of the diabetics person using association rule mining.