Hybrid deep neural network framework for scalable and robust cardiovascular disorder prediction

Abstract

Cardiovascular disorders remain the leading cause of global mortality, underscoring the urgent need for accurate and early prediction systems that can support timely medical intervention. Traditional diagnostic methods, while valuable, are often limited by subjectivity, time constraints, and the inability to fully capture complex patient data. To address these challenges, this study proposes a hybrid deep neural network (HDNN) framework that integrates convolutional neural networks (CNNs) and long short-term memory (LSTM) networks, supplemented with dense layers, to enhance predictive accuracy and robustness. The hybrid architecture leverages CNNs’ ability to extract spatial features and LSTMs’ strength in modeling temporal dependencies, thereby providing a comprehensive analysis of structured and unstructured patient data, including clinical records and health metrics. Rigorous preprocessing, feature selection, and domain specific knowledge integration further improve model performance compared to conventional machine learning (ML) approaches. Experimental evaluation on benchmark datasets, including the cleveland heart disease (HD) dataset and a combined multi-source dataset, demonstrated superior results, achieving accuracies of 98.86% and 97.52%, respectively. These findings highlight the potential of the HDNN framework to serve as a scalable, reliable, and clinically relevant tool, assisting healthcare professionals in early diagnosis, preventive care, and improved patient management.