Integrating blockchain and deep learning: a novel ensemble model for secure IoMT-driven intelligent healthcare solutions using ISSCNetV2 approach

Abstract

The Internet of Medical Things (IoMT) comprises a network of interconnected medical devices such as wearables, diagnostic tools, and implants that facilitate real-time data acquisition and remote healthcare monitoring. To ensure secure and reliable data transmission and storage in such environments, this study proposes an Enhanced Blockchain-based Intelligent Healthcare System with Ensemble Deep Learning (EBIHS-EDL). The system incorporates blockchain (BC) technology to maintain decentralized, tamper-proof records and employs a Bit-Level Chaotic Image Encryption Algorithm (BCIEA) for secure image encryption. Key generation is achieved using the Grasshopper–Black Hole Optimization (G–BHO) algorithm. To address the challenge of class imbalance in medical datasets, an Improved Tabular Generative Adversarial Network (ITGAN) is employed to synthesize minority class samples. For feature extraction, a Cross Siamese Res2Net (CSRes2Net) architecture is utilized, followed by classification using an integrated model, Improved ShuffleNetV2 and Spatiotemporal Convolutional Network-enhanced Transformer (ISSCNetV2). Comprehensive evaluations on benchmark medical datasets demonstrate the effectiveness of the proposed system, achieving an accuracy of 99.20%, sensitivity of 99.03%, and specificity of 99.46%. These results surpass those of existing models including DBN (94.15%), YOLO-GC (94.24%), ResNet (96.19%), VGG-19 (91.19%), and CDNN (95.29%), highlighting the superior performance and robustness of EBIHS-EDL in intelligent healthcare applications.