AN EFFICIENT METAHEURISTIC BASED RULE OPTIMIZATION OF APRIORI RARE ITEMSET MINING FOR ADVERSE DISEASE DIAGNOSIS MODEL

Abstract

In recent times, earlier identification and diagnosis of diseases have become an essential factor for reducing the death rate and severity of the diseases. Computational intelligence approaches such as rare association rule mining (ARM) are widely employed to diagnose the disease. This paper develops a novel krill herd algorithm based Apriori Rare itemset mining technique named KHAR-RIMfor adverse disease diagnosis. The presented KHAR-RIM model is mainly focused on three different diseases such as heart disease (HD),hepatitis, and breast cancer, with respect to the rare association rules.The presented model makes use of the Apriori Rare algorithm for generating the rare itemset rules. For improvising the outcomeof the performance of the rare itemset mining, the optimal rule generation of Apriori Rare takes place using the krill herd (KH) algorithm which is based on the herding nature of krills. The application of the KH algorithm into the rule generation process helps to attain maximum diagnostic outcomes. The performance of the KHAR-RIM model is validated using Cleveland HD, hepatitis, and Wisconsin Breast Cancer (WBC) dataset. The experimental values stated that the KHAR-RIM model has resulted in superior performance with the optimal rule generation of 12643, 152789, and 19821 rules on the applied using Cleveland HD, hepatitis, and WBC dataset respectively.