An Intelligent Adaptive Access Control Framework using Policy Gradient Reinforcement Learning

The adaptive access control mechanisms combined with policy gradient reinforcement learning can dynamically adjust the behavior of users in real time. The approach enables intelligent, situational decision-making in hospital information systems, which promotes their safety and effectiveness. The classical models of access control within healthcare are highly founded on the predetermined positions and set, rule-based permissions. These systems do not react to new threats and insider anabiosis, as well as unknown access patterns, and as such, are prone to incredibly dynamic clinical environments. In order to address these deficiencies, the Policy Gradient-based Dynamic Role Assignment framework applies reinforcement learning to continue maximizing role assignment. It relies on policy gradient techniques, through which access privileges are provided based on user actions, the context, and the past trend of the behavior in real-time. The Policy Gradient-based Dynamic Role Assignment is not responsive to risk scenarios by strictly separating roles. Nonetheless, it distributes them with dynamism as per the risks and needs available. An access decision making process that is context-sensitive and can increase the level of data security because only the authorized personnel can access sensitive patient data in the right circumstances. The beneficial effects of the supplementary policy entries on healthcare organizations are the heightened compliance of regulations, less breaches of policies, and enhanced access control. Past researches suggest adaptive role shifting within a dynamic healthcare context to be advantageous, improves the protection of confidentiality, and eases clinical practice. The dynamism of the conventional access control systems in clouds of healthcare information technology is developed into a strong and developed model which is known as the PG-DRA model.