An Efficient and Secure Blockchain-Based SVM Classification for a COVID-19 Healthcare System

Abstract

The recent coronavirus (COVID-19) pandemic has brought the world to an apocalyptic standstill with huge economic burden and catastrophic healthcare consequences. COVID-19 pandemic has neither clinically proven vaccine nor drugs so far. It is now obvious that the world is in dire need for non-clinical, real-time, faster and cost-effective and secure smart solutions for monitoring, contact tracing, and diagnosing/detecting COVID-19 patients and hence mitigating the burden on healthcare systems. Therefore, Machine learning approaches can be leveraged in all aspects that could impact the patients and the future care guidelines. However, ensuring the data privacy, security and the conformity to data protection regulations will become even more of a challenge. This leads to the requirement for secured and privacy-preserving machine learning mechanisms for COVID-19-based healthcare applications. In this paper, we present a blockchain-based privacy-preserving support vector machine (SVM) classification over vertically partitioned IoMT data for a clinical decision support (CDS) system. The proposed system does not require any intervention or direct interactions between data owners. Both local training and building the global classification model run on verifiable and private smart contracts rather than relying on untrusted third parties. We find that the proposed system is more secure and efficient.

Ähnliche Arbeiten