Frontiers in Health Informatics
Volume:3 Issue: 3, 2013

Since the most recent statistics report that in Iran and in the world, cardiovascular diseases are the main cause of mortality; most of researchers pay attention to these fields. Data mining, as one of the most important data analyzers, relationship explorers and event predictor tools, is an applicable tool in this research. One of the most important reasons of cardiovascular disease is hypertension. So in this research, a diagnosis method for the controlled/uncontrolled hypertension is presented. The input of the proposed method is the sequence of patients’ resorts to the doctor and the drugs which is prescribed to patients by the doctor. In this method, a sequence classification problem is changed to a regular one. In this research, to present patients’ resort, two feature vectors are explained and four famous classifiers are used. As a conclusion, the proposed method diagnoses the controlled/uncontrolled hypertension with the accuracy of 85% and furthermore the SVM classifier is the most efficient classifier between four famous used classifiers.

Growth of cardiovascular diseases and their effects on society and the high cost of imports, caused the medical community to pursue plans for further evaluation, prevention, early detection and effective treatment so valuable knowledge can be created by using data mining and knowledge discovery in cardiology centers that the discovered knowledge can improve the quality of service by medical center’s managers and It can also be used by doctors to treat and predict cardiovascular disease by their disease history. In this research, to improve the diagnosis of cardiac diseases a classification - feature selection approach has been proposed that were evaluated by using a data set of patient records of Amir-Al-Momenin hospital in Tehran. Our proposed method is selected by using the different ways of classification on existing data of healthy persons and patients to reach the most efficient way and highest accuracy. At the end the learned classifier are evaluated by a set of test specimen. Finally evaluation shows that the classification tree method C4.5 classification accuracy has earned the highest standard of accuracy rate as 95.8%.