Naïve Bayes based Summarizing Ruleset in Prediction of Diabetes Mellitus using Magnum Opus

Diabetes mellitus is the deficiency that is widely spreading nowadays. Manually diagnosing a person with diabetes is more complicated. If diabetes is not treated early it may lead to severe complications. We focus on Electronic Medical Records (EMR) to find out the factors that represent a patient with the risk of developing diabetes. We apply Apriori, Éclat and OPUS association rule mining techniques to generate the risk factors that occur frequently will help greatly in predicting diabetes. These frequent risk factors of each technique are subject to Naïve Bayes with which the chances for developing diabetes mellitus is predicted and the efficiency of each is obtained with respect to Success probability. In evaluating and comparing the previous techniques, OPUS is found to be efficient in predicting the factors that have a high risk of developing diabetes mellitus.