Internet of Medical Things Enabled Multimodal Framework: Deep Machine Learning for Chronic Cardiac Disease Prediction in Healthcare 5.0

Abstract

Accurate and early detection of chronic heart disease is vital, as it remains one of the leading global causes of mortality. Despite advancements in Smart Healthcare 5.0 and modern information technologies, reliable diagnosis of cardiovascular conditions remains a significant challenge. The Internet of Medical Things (IoMT) enables seamless data exchange between medical devices, supporting more precise and timely management of cardiac diseases. This study employs convolutional neural networks (CNNs) on electrocardiogram (ECG) image datasets to classify multiple heart conditions. The datasets include ECG scans labelled as Abnormal Heartbeat (ANHB), Myocardial Infarction (MI), History of Myocardial Infarction (HOMI), Atrioventricular Heart Block (AHB), COVID-19, Hypertrophic Cardiomyopathy (HMI), and Normal. A multimodal model integrating images of varying resolutions from two independent datasets was developed to improve classification performance. The proposed CNN model, trained and validated on preprocessed ECG images, achieved 97.18% training accuracy and 94.34% validation accuracy. By combining ECG data from diverse sources, the model enhances the identification of cardiac irregularities and provides a comprehensive diagnostic approach. This method demonstrates potential to support early detection, improve individualised treatment planning, and ultimately strengthen patient outcomes in managing chronic heart disease.