TWO PHASE PARALLEL FRAMEWORK FOR WEIGHTED COALESCE RULE MINING: A FAST HEART DISEASE AND BREAST CANCER PREDICTION PARADIGM

Abstract

Recently, there is an immense increase in the mortality rate of humans due to dangerous diseases, which is becoming a greater issue across the globe. The only solution to this issue is the early detection of infectious diseases, so that the seriousness of their symptoms can be reduced before reaching an adverse stage. In recent days, associative rule mining, which is a computational insight strategy is being more commonly utilized for early risk prediction of the disease. In the case of rule mining, there is a massive count of the frequent patterns that might deviate from the detection mechanism. Therefore, different customized algorithms are being implemented. Among them, the Apriori algorithm is a standardized model which is good in detecting the more frequent patterns. But, owing to a huge count of candidates as well as scans of the database, the ties technique has become inefficient. Therefore, to override these issues and to find a promising solution for the early disease prediction, “a new 2-phase parallel processing based Coalesce based Binary (CBB) Table” is introduced in this paper. The proposed disease prediction model involves: pre-processing, 2-phase parallel processing, weighted coalesces rule generation, optimal feature extraction, and classification. Particularly, for selecting the optimal features, the Grey Wolf Levy update – dragonfly algorithm (GWU–DA) algorithm is used and a hybrid classification model that incorporates “Support vector Machine (SVM) and Deep Belief Network (DBN)” is used to predict the presence of disease. Finally, the validation of this work over the extant models is accomplished in terms of various performance measures.