Hybrid model with improved score level fusion for heart disease classification

The heart, an essential organ of the body, pumps blood filled with oxygen to every area of the body through a complex network of veins and arteries. Any condition affecting the heart is referred to as heart disease. It is essential to start the right treatments early on in order to reduce the death toll. Predicting cardiac disease before heart attacks or strokes is therefore a significant challenge for the healthcare industry. This article introduces a methodology for predicting heart disease which consists of the three steps: (1) pre-processing, (2) feature extraction, and (3) disease prediction. Initially, the input data is pre-processed using improved data normalization. Next, the features are extracted from the normalized data; they are higher-order statistical features, information gain-based features, mutual information-based features, and improved entropy-based features. The results are then decided utilising an improved score-level fusion procedure, which is used to the suggested hybrid classification model that incorporates models like Deep Maxout and QNN algorithms. By adjusting the Deep Maxout and QNN classifiers' ideal weights, a self-improved Beluga whale optimisation is developed for optimal training to increase the performance of the hybrid model. When using dataset 1, the Hybrid + SI-BWO produced a precision of 94.25%, while CNN, DBN, Bi-LSTM, SVM, RNN, DL, and RF all produced precisions of 75.04%, 77.38%, 80.17%, and 86.52%, respectively.