Rule Discovery for Diabetes Mellitus Diagnosis using Ant-Miner Algorithm

摘要

Diabetes mellitus currently affects over 425 million people worldwide. According to the WHO report, by 2045 this number is expected to rise to over 629 million. The disease has been named the 2nd of NCDs (Non-Communicable diseases) in Thailand. In diagnosis of Diabetes mellitus are done mostly by expertise and experienced doctors, but still there are cases of wrong diagnosis. Patient have to undergo various test which are very costly and sometimes all of them are not required so in this way it will hugely increase the bill of a patient unnecessarily. This paper presents diabetes mellitus diagnosis system by analyzing the patterns via Pima Indian Diabetes Dataset (PIDD). The system is composed of main process, Pima Indian Diabetes Dataset are cleaned and transformation. Normal distributions are employed by Z-transform function. In rule discovery for diagnosis, we used the Ant-Miner classifier to classify Diabetes by assuming that the feature is features Diagnosis. This experiment, Ant-Miner algorithm is adapted, with a small change to increase the accuracy rate. The result of this experiment is more than 86% accuracy rate and shows that the constructed data mining model could assist health care providers to make better clinical decisions in identifying diabetic patients.