Diagnosis of coronary artery disease using an efficient hash table based closed frequent itemsets mining

This paper proposes an efficient hash table based closed frequent itemsets (HCFI) mining algorithm to envisage coronary artery disease early. HCFI algorithm generates closed frequent itemsets efficiently by performing intersection operation on transaction id’s of itemset without considering the name of item/itemset. The employed hash table reduces search efficiency to O(1) or constant time. HCFI algorithm is applied on the UCI (University of California, Irvine) Cleveland dataset, a biological database of cardiovascular disease to generate closed frequent itemsets on the dataset. The findings of HCFI algorithm are (1) it determines a set of distinguished features to differentiate a ‘healthy’ and a ‘sick’ class. The features such as heart status being normal, oldpeak being less than or equal to 1.2, slope being up, number of vessels colored being zero, absence of exercise-induced angina, maximum heart rate achieved between 151 and 180 are referred as ‘healthy’ class. The features like chest pain are being asymptomatic, heart-status being reversible defect, slope being flat, and presence of exercise-induced-angina and serum cholesterol being greater than 240 indicate a presumption of heart disease to both genders. (2) It predicts that females have less chance of coronary heart disease than males. This algorithm is also compared with two other state-of-the-art-algorithms ‘NAFCP’ (N-list based algorithm for mining frequent closed patterns) and ‘PredictiveApriori’ to show the effectiveness of the proposed algorithm.