Heart disease prediction using hybrid optimization enabled deep learning network with spark architecture

Abstract

Analyzing massive amounts of data that contain many sorts of data is known as big data analytics. Additionally, the bulk of applications in the actual world need a significant amount of information. Machine learning techniques are used to automatically identify the types and severity of cardiac disease due to the rapid increase of biomedical and healthcare information. However, the ML approach has a number of drawbacks, and because of the complexity of the material, it did not always produce the best results. As a result, an improved deep learning strategy offers a superior fix for this problem. In this study, a brand-new method for diagnosing heart disease—the Grey Wolf Horse Herd optimization-based Shepard Convolutional Neural Network is developed. Here, the master node and slave node-based Spark architecture is used to carry out the heart disease detection process. Preprocessing and feature fusion are carried out in the slave node, whilst heart disease detection is done in the master node. Z-score normalization and missing value imputation are used in this case for pre-processing. The feature fusion is then carried out utilizing Hellinger distance and Deep Q Network (DQN). Furthermore, the ShCNN, which was trained using the created GWHHO algorithm, is used to identify heart disease. The Grey Wolf Optimizer (GWO) and Horse Herd Optimization (HHO) algorithms are also incorporated into the newly developed GWHHO algorithm. Additionally, employing the VA Long Beach dataset, the experimental of the developed model yields improved results in terms of testing accuracy, sensitivity, and specificity of 0.9325, 0.9472, and 0.9142.