Risk and UCON-based Access Control Model for Healthcare Big Data

Abstract

Abstract The rapid development of healthcare big data has brought certain convenience to medical research and health management, but privacy protection of healthcare big data is an issue that must be considered in the process of data application. Access control is one of the methods for privacy protection, but traditional access control models cannot adapt to the dynamic, continuous, and real-time characteristics of healthcare big data scenarios. In this paper, we propose an access control model based on risk quantification and usage control (RQ-UCON). The model adds a risk quantification module to the traditional UCON model to achieve privacy protection of medical data. This module classifies risks into direct and indirect risks and quantifies them based on the physician's visit history. The model stores the quantified risk values as subject attributes. The RQ-UCON model uses an improved Exponentially Weighted Moving Average (EWMA) and penalty factors to predict risk value and to update the risk values of the subject attributes in real-time. The RQ-UCON model uses agglomerative hierarchical clustering to cluster the risk values of physicians within the department, resulting in risk intervals for each physician's operational behavior. Each risk interval is stored as a condition in the RQ-UCON model. Finally, according to the model whether the subject attributes meet the model conditions to determine whether the subject has the corresponding access rights, and according to the risk interval to grant the subject the corresponding access rights. Through the final experiment, it can be seen that the access control model proposed in this paper has a certain control on the excessive access behavior of doctors and has a certain limitation on the privacy leakage of healthcare big data.