Enhancing Data Diversity and Traceability to Mitigate Bias in Healthcare AI: A Blockchain-Driven Approach

Abstract

In the field of healthcare AI, the prevalence of algorithmic bias poses a significant challenge, often stemming from data biases. This research presents a novel approach to mitigate data biases by harnessing the potential of blockchain technology to enhance data diversity and traceability within federated learning systems. Our focus is to establish a framework where blockchain bolsters data integrity and broadens data inclusivity. The core of our methodology lies in leveraging blockchain's distributed architecture, which inherently facilitates the integration of a varied and voluminous data pool. This contrasts sharply with limitations found in conventional health data standards like HL7 and FHIR, which often lack in providing a comprehensive data spectrum. Our study contributes to the ongoing discourse on fair and unbiased AI development in healthcare.