Darwinian Lion Swarm Optimization-Based Extreme Learning Machine with Adaptive Weighted Smote for Heart Disease Prediction

Predicting Cardiovascular Diseases (CVD) is essential in reducing the deaths associated with different heart diseases. However, there are complexity and class imbalance problems in existing methods that degrade the overall effectiveness. Therefore, advanced frameworks are essential for heart disease detection models. This paper proposes a competent heart disease recognition framework using proficient machine learning (ML)-based methods for each clinical data analysis stage. Initially, the pre-processing stage introduces three methods for enhancing the input clinical data quality. Firstly, the Hierarchical Density-based Spatial Clustering of Applications with Noise based on the Neighbor Similarity (HDBSCAN-NS) is proposed to eliminate the data outliers without increasing the complexity. Secondly, Class Median based Missing Value Imputation (CMMVI) approach is proposed to impute the missing values based on the class distribution. Thirdly, Adaptive Weighted Synthetic Minority Oversampling Technique with Natural Neighbors (AWSMOTE-NN) is proposed to resolve the drawbacks of collinearity in the imbalanced dataset and improve the class balancing. After pre-processing, the features are extracted, and optimal feature subsets are selected using the Owl Optimization algorithm (OOA). These selected features are used by the proposed hyper-parameter optimized classifier of Darwinian Lion Swarm Optimization-based Extreme Learning Machine (DLSO-ELM) is utilized for detecting heart disease. UCI heart disease datasets are used to validate the proposed framework, and the results showed that the OOA-DLSO-ELM-based approach provides better heart disease prediction with high accuracy and low complexity.