Protecting health data in the cloud through steganography: A table-driven, blind method using neural networks and bit-shuffling algorithm

Abstract

Securing health data stored in the cloud, including DNA sequences, patient information, and treatment histories, has become increasingly difficult due to modern cyber threats. To address this challenge, one approach is to employ steganography, a method of concealing sensitive information by merging it with other data. Recent studies have explored hiding secret information within DNA sequences, but these approaches have drawbacks. They either fail to generate a suitable cover from the modified sequence, known as the stego DNA sequence, or result in a significant increase in data that needs to be added to the original DNA sequence. Some methods even inadvertently include the original secret information or cover within the stego DNA sequence, creating security risks. To overcome these limitations, this article presents a table-driven, blind steganographic technique specifically designed for DNA sequences stored in the cloud. Our proposed technique offers superior embedding capabilities, reduces DNA expansion, and minimizes security risks. The proposed method introduces two unique proposals: Employing machine learning techniques for secret encryption and utilizing bit-shuffling to enhance message protection, thereby increasing resistance against cyber-attacks. Furthermore, the method supports hiding various types of messages. Compared to existing methods, the suggested scheme achieves a lower stego expansion rate. Additionally, a newly introduced parity check method strengthens security by preventing modifications to the stego DNA, such as man-in-the-middle attacks, chosen stego attacks, and modification attacks. These innovations enhance the overall security capabilities of the system. Experimental results validate the effectiveness of the proposed method, demonstrating accurate and error-free reconstruction of the cover, healthcare data, and DNA sequence. The proposed technique outperforms competing methods across all performance metrics, establishing its superiority in securing health data in the cloud.