A Machine Learning based Scalable Blockchain architecture for a secure Healthcare system

Abstract

The evolution of the Industrial revolution from 3.0 to 4.0 has transformed the Healthcare environment. Patient Electronic Health Records (EHR) are shared with medical research institutes for clinical research and to manage national disease outbreaks. Healthcare systems implementing centralized machine learning models risk cyberattacks exposing private patient data. Blockchain-based data storage systems enable data security of EHR. However, the low transactions/minute of decentralized systems limit the performance of Healthcare systems and increase network bottleneck concerns. In this paper, we propose a Machine Learning based Blockchain architecture for secure Healthcare systems to preserve patient data privacy using Federated Learning and address Blockchain bottleneck issues by adding sidechains for processing growing transaction requests. A local model using machine learning trains data locally in hospitals and uploads it via Smart Contracts to the Public Healthcare System for global model training. Sidechains enable increased processing speed of Smart Contracts reducing congestions in the network and increasing the transactions per second in the mainchain.