Automated heart disease prediction model by hybrid heuristic-based feature optimization and enhanced clustering

The main intent of this paper is to implement a novel clustering model for heart disease prediction with numerical data and ECG signals using an optimal feature extraction approach. Rather than the direct use of numerical data to clustering, the Electrocardiogram (ECG) signals are initially subjected for the signal decomposition using Discrete Wavelet Transform (DWT), and dimensionality reduction is performed through Principal Component Analysis (PCA). Both the data are processed for the optimized feature extraction stage. Here, the hybrid meta-heuristic concept is adopted for the optimized feature extraction based on Jaya Algorithm with Red Deer Algorithm (J-RDA). Once the feature optimization is done, the hybrid clustering is formed by integrating the optimized Density-based Spatial Clustering of Applications with Noise (DBSCAN) and optimized K-Means Clustering (KMC), in which the proposed J-RDA is used for tuning the significant parameters. Moreover, the objective model for feature optimization and optimized hybrid clustering for proposed heart disease prediction tries to solve the multi-objective function. The results reveal that the proposed model achieves good performance in rectifying the problems in heart disease prediction for dual data types.