Sci Rep. 2026 Jun 11. doi: 10.1038/s41598-026-57620-0. Online ahead of print.
ABSTRACT
Cardiovascular diseases are a primary global health concern, requiring continuous monitoring and accurate diagnostic mechanisms for early detection and prevention. The integration of the Internet of Medical Things (IoMT), deep learning, and blockchain technologies has enabled intelligent healthcare systems capable of real-time cardiac assessment and enhanced security in the management of medical information. However, existing heart disease monitoring systems are affected by noisy physiological signals, inefficient feature extraction, limited classification accuracy, and insufficient security in distributed environments, motivating the development of a robust and reliable diagnostic framework with improved security. To address these challenges, this research proposes an IoMT and Blockchain-Based Heart Disease Monitoring System Using Hybrid Black-Winged Spherical Structural Graph Convolution Neural Network (HBW-SSG-CNN). The proposed workflow begins with IoMT-based acquisition of ECG and PCG signals, followed by preprocessing using quasi-cross bilateral filtering (QCBF) to suppress noise while preserving critical cardiac structures. Signal decomposition is performed using Spectral Envelope-Based Adaptive Empirical Fourier Decomposition (SE-AEFD), followed by feature extraction using the Short-Time Quaternion Quadratic Phase Fourier Transform (ST-QQPFT). Feature dimensionality is optimized using the Success-Based Optimization Algorithm (SBOA), and heart disease classification is performed using a Hybrid Structural Graph Attention Network with Spherical Convolutional Neural Network (HS-GAT-SCNN), further enhanced by the Black-Winged Kite Algorithm (BWKA). To enhance data integrity and improve security, an Adaptive Hash Algorithm with Weighted Probability Model (AHA-WPM) is integrated with a Fair Consensus Blockchain for Heterogeneous Miners (FCB-HM). The proposed model is evaluated using publicly available benchmark datasets, including PhysioNet cardiac signal datasets and the Cleveland heart disease dataset. Experimental evaluation demonstrates superior performance with an accuracy of 99.21%, sensitivity of 98.94%, specificity of 99.08%, and F1-score of 99.02%. The results confirm that the proposed framework provides a highly accurate, scalable, and security-enhanced solution for near real-time heart disease monitoring in IoMT-enabled healthcare systems.
PMID:42277292 | DOI:10.1038/s41598-026-57620-0