Privacy Parameter Setting and Usability Optimization Algorithm for Medical Data

In the era of the Consumer Internet of Things (CIoT), the processes of data publication, querying, and sharing in healthcare play a crucial role in protecting patient privacy. However, existing differential privacy techniques often suffer from issues such as unreasonable privacy parameter allocation and low data utility. To address these challenges, this paper proposes a parameterization and data optimization algorithm based on differential privacy for the healthcare data publication environment. The algorithm divides data users into different access identities and maps quantified user credibility and the data sensitivity of the accessed dataset to the privacy parameter $\epsilon $ . This parameter is then applied to the Laplace mechanism for adding noise to the healthcare data. Additionally, this paper improves the $(\alpha, \beta)$ -useful mechanism by associating the optimized parameters $\alpha $ and $\beta $ with user credibility, and employs this mechanism to optimize the noisy data, thereby enhancing data utility. Experimental analysis demonstrates that the proposed algorithm achieves a reasonable allocation of privacy budget, assigning privacy parameters suitable for different user identities. It effectively protects privacy data to a certain extent while improving the utility of healthcare data, providing users with medical data of varying precision according to their respective identities.