A KNN Approach for Blockchain Based Electronic Health Record Analysis

Abstract

AbstractBlockchain plays a significant role to connect end users with utmost security and reliability as it is well-suited to handle critical privacy issues in medical and business data sharing. It is considered as one of the best cutting edge technology in providing a suitable framework for sharing critical medical information across various experts globally. In this paper, an integrated method is designed to combine Blockchain with data analysis for an automated and secure sharing of medical datasets among healthcare experts in improving their patient diagnosis and advice. The first key component of our framework involves classification of a large number of medical records containing hundreds of variables or features using linear discriminant analysis (LDA) and K-nearest neighbours (KNN) algorithm. Using LDA, highly related and significant features are grouped for sharing them with medical experts in order to quickly and accurately reach a decision. The next key component of our framework addresses the challenges involved in sending the processed information to the targeted persons by introducing a dynamic Blockchain discovery process. For this, we develop an automated Blockchain discovery system that repeatedly checks the nodes in the Blockchain to transfer the information in order to ensure robust security. Further, we evaluate the effectiveness of KNN algorithm with and without LDA in making accurate decisions. A rigorous experimental analysis confirms that LDA based KNN algorithm is preferable in this context.