Blockchain enabled deep learning model with modified coati optimization for sustainable healthcare disease detection...

The growing number of patients and the emergence of new symptoms and diseases make health monitoring and assessment increasingly complex for medical staff and hospitals. The execution of big and heterogeneous data gathered by medical sensors and the necessity of patient classification and disease analysis have become serious problems for various health-based sensing applications. The significant features of healthcare are the privacy of medical details and the accuracy of disease identification. One of the key benefits of the healthcare system is the ability to predict diseases early. Recently, the progress of artificial intelligence (AI) in the healthcare system has been a high priority. Machine learning (ML) and deep learning (DL) effectively make analyses and strategic decisions for the healthcare system. This manuscript proposes a Modified Coati Optimization Driven Blockchain for Healthcare Disease Detection and Classification (MCODBC-HDDC) method. The presented MCOBC-HDDC method provides an efficient and accurate disease diagnosis, utilizing a system that depends on DL techniques. Initially, the MCODBC-HDDC method incorporates BC technology to ensure secure data sharing and management, providing a decentralized and tamper-proof environment for patient data. In the data preprocessing stage, the MCODBC-HDDC model employs Z-score normalization to standardize the data and improve performance. For the optimal subset of features, the spotted hyena optimization algorithm (SHOA) model is used. Furthermore, the attention bidirectional gated recurrent unit (ABiGRU) method is implemented for disease detection and classification. Finally, the hyperparameter selection of the ABiGRU method is performed by utilizing the modified coati optimization algorithm (MCOA) method. The experimental analysis of the MCODBC-HDDC approach is examined under the HD dataset. The performance validation of the MCODBC-HDDC approach portrayed a superior accuracy value of 97.36% over existing models.