A Novel Data Protection Model for Secure Medical Data in Cloud Environments Using Scaling Wide Residual Networks Based Key Generation

The Personal Health Record (PHR) system manages their medical records in the cloud independently by enabling them to upload and control access to sensitive health data. To address the security, a novel protection model based on key generation and Attribute‐Based Encryption (ABE), named Scaling Wide Residual Networks with Fractional Gazelle Brown Bear Optimization Algorithm (SWideResNet_FGez‐BOA) is introduced in this research. The SWideResNet_FGez‐BOA operates in two phases: PHR upload to the cloud and PHR retrieval from the cloud. In the first phase, patients, who are known as data owners, create accounts to manage their PHR files. Before uploading, the PHR files undergo a robust ABE‐based encryption process using Chebyshev polynomials for cloud security. SWideResNet_FGez‐BOA, which is the combination of Scaling Wide Residual Networks (SWideResNet) and an FGez‐BOA, is used for generating the keys. The FGez‐BOA is the integration of Fractional Calculus (FC), Gazelle Optimization Algorithm (GOA), and Brown Bear Optimization Algorithm (BOA). In the second phase, the PHR retrieves the uploaded data from the cloud. The SWideResNet_FGez‐BOA demonstrated a key generation time of 11.765 ms, an encryption time of 208.765 ms, a decryption time of 195.765 ms, a memory of 1.865 MB, a conditional privacy of 0.979, and a normalized variance of 0.033.