APDRChain: ANN Based Predictive Analysis of Diseases and Report Sharing Through Blockchain

Abstract

A huge amount of healthcare data (structured and unstructured) is currently available to medical specialists, indicating details of clinical symptoms. Each type of data provides information that must be properly analyzed for healthcare diagnosis. To simplify the diagnostic process, avoid misdiagnosis as well as early detection, artificial intelligence (AI) that aims to mimic human cognitive functions may be employed. Current AI techniques that are used for structured data include machine learning methods, such as the classical support vector machine, artificial neural network, and the modern deep learning. Natural language processing is mainly used for unstructured data. In this paper, we have adopted artificial neural network by using adaptive learning algorithms to handle diverse types of cardiovascular clinical data and integrate them into categorized major cardiovascular disease outputs such as heart failure, aortic aneurysm, cardiomyopathy, cerebrovascular disease, etc. These outputs are then shared as reports to patients as well as doctors by an efficient report sharing scheme called APDRChain, which combines blockchain and structured peer-to-peer network techniques with clever cryptography to create a consensus mechanism. The evaluation results show that APDRChain can achieve higher efficiency and satisfy the security requirements in report sharing.