Medical Data Mining using Fuzzy Expert System and Imperialist Competitive Algorithm

Huge amount of data in medical fields is pushing the experts to the using of data mining techniques. There are many factors which effect on the risk of a disease, considering all of them at a time is difficult for a physician. Data mining techniques and machine learning algorithms are tools make it easier to diagnose diseases. In this article, for the first time a new approach, which uses fuzzy expert system Imperialist Competitive Algorithm (ICA), is proposed. The proposed system uses ICA to tuning the membership functions. The advantage of the system in comparison with similar researches is the interpretability of decisions. Interpretation of the rules is determined by the length and number of rules employed. The system gained acceptable results by 11 rules and the average length of 4.3. The proposed approach tested on three disease datasets, Coronary Artery Disease (CAD), Hepatitis and Diabetes and gained acceptable classification accuracies in comparison with similar works. To evaluate ICA, another evolutionary algorithm had tested on datasets, which is PSO.

Keywords:

ICA , Fuzzy expert system , Decision tree , Membership function , CAD , Diabetes , Hepatitis , PSO