Heart Disease Prediction Model Using Varied Classifiers with Score-Level Fusion

Abstract

This paper aims to introduce a novel heart disease prediction model. Originally, the input data is subjected for preprocessing, in which the data cleaning takes place. The features like statistical, higher order statistical features, and symmetrical uncertainty are extracted from the preprocessed data. Then, the selected features are subjected to the classification process with an ensemble model that combines the classifiers like deep belief network (DBN), random forest (RF), and neural network (NN). At last, the score level fusion is carried out to provide the final output. To make the classification more precise and accurate, it is intended to tune the weights of DBN more optimally. A new self-adaptive honey bee mating optimization (SAHBMO) algorithm is implemented in this work for this optimal tuning. Finally, the performance of the presented scheme is computed over the existing approaches in terms of different metrics.