Blockchain-based decentralized smart healthcare using improved wild horse optimizer with Graph Convolutional Autoencoder in IoT environment

The Internet of Things (IoT) continues to expand by incorporating physical devices, software, computing systems, and hardware that facilitate communication and data exchange. Its integration into healthcare, specifically in the realm of smart healthcare, has contributed significantly to the rise of big data within the medical field. The adoption of IoT-enabled wearable technologies by healthcare professionals aims to streamline diagnosis and treatment processes. However, security and privacy concerns associated with data storage and transmission pose significant challenges to the efficacy and trustworthiness of these systems. To address these concerns, this article presents a blockchain-assisted centralized smart healthcare framework, which utilizes the Improved Wild Horse Optimizer (IWHO) and Graph Convolutional Autoencoder (GCAE) in an IoT environment to ensure secure and accurate disease detection. The BIWHO-GCAE framework consists of three main components: Inception v3-based feature extraction, IWHO-based hyperparameter tuning, and GCAE-based classification. The experimental evaluation, conducted using the benchmark skin lesion dataset, shows that the BIWHO-GCAE method outperforms current state-of-the-art deep learning models, demonstrating improvements of 2.62% in accuracy, 3.07% in sensitivity, and 7.28% in specificity. These results highlight the potential of the BIWHO-GCAE framework to enhance diagnostic performance while ensuring the security and privacy of healthcare data in decentralized IoT-based systems. • Development of BIWHO-GCAE Technique integrating BC technology to enhance security and privacy in the healthcare domain, especially for IoT devices that collect sensitive health data. • The research details a three-stage disease detection module that includes Inception v3 feature extraction, IWHO-based parameter tuning, and GCAE-based classification. This approach allows for efficient and accurate disease diagnosis by optimizing the selection of hyperparameters and improving the classification performance of Deep Learning (DL) models. • Evaluation of the BIWHO-GCAE technique using a benchmark skin lesion database demonstrates the effectiveness of the method in various performance metrics. These results highlight the ability of the proposed system to accurately classify different types of skin lesions, thereby underscoring its potential utility in real-world medical applications.