An efficient hybrid optimization algorithm for detecting heart disease using adaptive stacked residual convolutional neural networks

Abstract

Heart disease remains a leading global cause of mortality. Cardiovascular disease is the most common vascular disorder contributing to heart-related issues. Critical factors impacting cardiovascular health include blood pressure, glucose control, and cholesterol management. The escalating mortality rates associated with heart disease underscore the urgency for precise and secure diagnoses to facilitate timely intervention and prevention. Incorrect diagnoses of cardiovascular disease can have devastating consequences, highlighting the critical need for accurate assessments. In this study, we introduce an intelligent system for heart disease prediction based on the integration of the ACLS-RCNN and ICSOA techniques. The workflow begins with the collection and preprocessing of patient information. Data preprocessing focuses on addressing missing values, employing IQR-RS scaling, and handling imbalanced data using the ROS technique. Next, features are extracted from the pre-processed data utilizing kernel-based linear discriminate analysis. To reduce the dimensionality of the dataset, hybrid optimization (CI-AO + GI-SMO) is employed to select informative attributes. Disease prognoses are then generated through the application of the ACLS-RCNN model, with further enhancement achieved by the Improved Cuttlefish-Swarm Optimisation Algorithm (ICSOA). Through extensive Python-based simulations, we evaluate the recommended system's performance in terms of accuracy, precision, recall, and f1-score, comparing it against established methodologies. Our software demonstrates remarkable accuracy in clinical data prediction and diagnosis, offering a promising tool for enhancing heart disease prognosis and early detection.