Optimal Key-Aided Elliptic Galois Encryption for Secured Data Transmission in IoT With Adaptive and Sparse Attention-Based Dense Bi-GRU to Predict the Diseases in Healthcare Sector

The transparent framework of IoT systems makes them susceptible to a range of digital risks, most notably side-channel attacks, eavesdropping, and Denial of Service (DoS). These attacks affect the medical data when it is in transit (during transmission), in use (during processing), and at rest (stored on cloud platforms). To enhance disease prediction on IoT platforms, this research implements a specialized deep learning-based architecture. In the developed framework, a secure data transmission model utilizing the Optimal Key-based Elliptic Galois Encryption (OK-EGE) technique is employed to secure the “in transit” stage, ensuring confidentiality during transmission. Next, the secured data is considered to execute disease prediction in the IoT network. In this phase, disease prediction is executed using Adaptive and Sparse attention-based Dense Bidirectional Gated Recurrent Unit (ASDB-GRU). It mitigates risks during processing and storage for the “in use” and potentially “at rest” stages. Later, several parameters in ASDB-GRU are tuned using an improved optimization technique named Revised Arbitrary Variable-based American Zebra Optimization Algorithm (RAV-AZOA). Finally, disease-predicted outcomes are obtained from the developed ASDB-GRU. The developed RAV-AZOA-SDB-GRU model achieved a superior accuracy of 97.13%, which is better than existing methodologies.