A New ICA-Based Algorithm for Diagnosis of Coronary Artery Disease

Large amount of data in recent years have pushed experts to use data mining techniques in all fields. Data mining is a process in which useful information from raw data is obtained; it can also be used in classification problems. Lately, in some systems, especially in medical systems, experts have tried to combine data mining techniques and evolutionary algorithms to get accurate results. One of the most critical diseases, which has a considerable mortality rate in the world, is coronary artery disease. To improve the diagnosis of this dangerous disease in the early stages, we proposed a system which uses data mining techniques and an evolutionary algorithm called Imperialist Competitive algorithm (ICA). The proposed system used an algorithm based on the decision tree to reduce the data dimension and to produce valid rules. Then, a fuzzy system is created. Tuning fuzzy membership functions were done using ICA and Improved ICA to optimize the results. Since the convergence speed is one of the important factors in an evolutionary algorithm, a change was made in this algorithm so that the convergence occurs more quickly. The results show that ICA and Improved ICA produce the same results in classification accuracy, but the convergence time is different. The proposed system gets an accuracy of 94.92 %, which is high in comparison with similar works.