E-Health Blockchain: Conception of a New Smart Healthcare Architecture Based on Deep Reinforcement Learning

Abstract

The recording, manipulation and exploitation of medical data represent a challenge for the protection of medical data against malicious users. Several researchers have proposed protocols and architectures to assure the integrity, confidentiality, and privacy of data storing on Cloud. Subsequently, the use of Blockchain and especially hybrid is the best promise to revolutionize the healthcare industry to store data securely. In this article, we inroduce a smart conception of healthcare architecture based on deep reinforcement learning, called E-health Blockchain. It combines three phases, from the patient's registration at the hospital to his recovery: (1) the pre-processing phase where the patient starts depositing his data, (2) the processing phase by the service designated by deep reinforcement learning and (3) the control phase where the doctor decides on the basis of certain information and analyses whether the patient can be discharged or not. We start the first phase by initializing the Blockchain, where we define the different blocks that make up the hybrid Blockchain, then the symptoms stored by the patients are classified using reinforcement learning to determine the appropriate service for each patient. In the second phase, we process the patient's condition in order to prescribe an appropriate diagnosis for their illness, or even a more in-depth consultation in the case of a complicated pathology. In the final phase, a decision is taken on the basis of the conditions for recovery met by the patient. This architecture makes it possible to achieve greater efficiency and meet security requirements in terms of confidentiality and secure sharing of patient data, on the one hand, and ubiquitous maintenance in the event of failure of a given service, on the other.