A Novel Optimization Enabled Ensemble Deep Learning Model for Heart Disease Detection

Background Cardiovascular diseases continue to represent one of the predominant causes of global mortality. Reducing mortality rates requires prompt detection of these conditions. The development of complex predictive models is required because traditional methods for the detection of heart disease frequently show limitations in accuracy and robustness. Objective This study presents an enhanced ensemble method based on deep learning for heart disease detection, which uses a novel training algorithm, Accelerated Sparrow Search Algorithm (ASSA), to increase detection precision. Methods Data pre‐processing, feature selection, and ensemble‐based classification are the three main stages of the suggested framework. To address data inconsistencies, min‐max normalization is used as part of the data pre‐processing. The feature selection stage uses a hybrid approach that combines information gain and wrapper techniques to extract features as best as possible. Recurrent Neural Network (RNN), Deep Residual Network (DRN), and Deep Belief Network (DBN) are all integrated in the final classification phase. The parameters of the networks are carefully refined using the ASSA optimization technique. Results The experimental analysis confirms that the ASSA‐refined ensemble approach outperforms conventional methods in terms of precision, sensitivity, and specificity. The model attained an accuracy of 95%, sensitivity of 97%, and specificity of 94%, thereby underscoring its superior efficacy in the detection of heart disease. Conclusion Compared to existing methods, the suggested ensemble approach is supported by ASSA optimization, and deep learning shows improved capabilities for heart disease detection.