Reinforcement learning approaches for efficient and secure blockchain-powered smart health systems

Abstract Emerging technological innovation toward e-Health transition is a worldwide priority for ensuring people’s quality of life. Hence, secure exchange and analysis of medical data amongst diverse organizations would increase the efficiency of e-Health systems toward elevating medical phenomena such as outbreaks and acute patients’ disorders. However, medical data exchange is challenging since issues, such as privacy, security, and latency may arise. Thus, this paper introduces Healthchain-RL, an adaptive, intelligent, consortium, and secure Blockchain-powered health system employing artificial intelligence, especially Deep Reinforcement Learning (DRL). Blockchain and DRL technologies show their robust performance in different fields, including healthcare systems. The proposed Healthchain-RL framework aggregates heterogeneous healthcare organizations with different requirements using the power of Blockchain while maintaining an optimized framework via an online intelligent decision-making RL algorithm. Hence, an intelligent Blockchain Manager (BM) was proposed based on the DRL, mainly Deep Q-Learning and it is variations, to optimizes the Blockchain network’s behavior in real-time while considering medical data requirements, such as urgency and security levels. The proposed BM works toward intelligently changing the blockchain configuration while optimizing the trade-off between security, latency, and cost. The optimization model is formulated as a Markov Decision Process (MDP) and solved effectively using three RL-based techniques. These three techniques are Deep Q-Networks (DQN), Double Deep Q-Networks (DDQN), and Dueling Double Deep Q-Networks (D3QN). Finally, a comprehensive comparison is conducted between the proposed techniques and two heuristic approaches. The proposed strategies converge in real-time adaptivity to the system status while maintaining maximum security and minimum latency and cost.