Heart Disease Prediction Using Parallel Ensemble Deep Learning Techniques

Abstract

To guarantee proper categorization and allow cardiac specialists to treat patients effectively, it is essential that heart disease be diagnosed and prognoses. Implementation of machine learning in the medical field have expanded as a result of its capacity to recognize patterns in data. The objective of this research is to decrease the amount of deaths from heart disease by creating a model that can accurately forecast these issues. The five primary stages of this study are preprocessing, clustering, extraction of feature, feature selection, and classification. Initially the min-max normalization is used for pre-processing. The K-Means Clustering (KMC) technique is used for clustering with the goal of improving classifier accuracy. This work suggests using a Principal Component Analysis (PCA) to efficiently extract features. After that, the feature selection process is finished using the modified whale optimization algorithm method. To choose the most pertinent and significant characteristics from the cardiovascular illness dataset, it computes the optimal fitness value. Finally, the classification is done using parallel ensemble deep learning model as VGG16, Inceptionv2, ResNet50 and used the Mutation Albatross Optimization Algorithm (MAO) for hyperparameter optimization.  To improve health outcomes and lessen the demand on healthcare, these models can be used to create future cardiovascular disease detection systems.