Data Sharing as a Debiasing Measure for AI Systems in Healthcare: New Legal Basis

Abstract

Healthcare data sharing is increasing through platforms devoted to the collection of personal health data; due to the sensitive nature of healthcare data, and the sensitivity of personal data in data protection law, ethical challenges are posed for the disclosure and sharing of healthcare data even in scientific research. However, if data sharing can be implemented as a debiasing measure to improve Machine Learning (ML) systems used in healthcare, the sharing and processing of sensitive healthcare data might pose as a solution most stakeholders seek. The questions regarding data sharing involve the balance we need, between protection of privacy and improving systems from biases; in other words, balancing individual versus collective rights. After the proposal for a regulation on the European Health Data Space (EHDS), the new legal basis for data sharing in the recently approved Data Governance Act (DGA) followed by the Data Act proposal, and with the provisions in the Artificial Intelligence Act (AIA), data sharing might take a different turn. This article examines the new legal basis for data sharing as a debiasing measure to improve Artificial Intelligence (AI) systems in healthcare.