Technical and Legal Aspects Relating to the (Re)Use of Health Data When Repurposing Machine Learning Models in the EU

The recent wave of data driven technologies such as artificial
intelligence (AI) and the internet of things (IOT), to name a few, opened
up many new horizons and placed data at the heart of technological in-
novation. Despite their promising results in various domains, these tools
are data greedy in nature which add to their need for more datasets.
Luckily, in technologies like machine learning, there is a possibility to
reuse the existing machine learning models, also known as ’knowledge
transfer’ for other tasks. However, this solution when examined under
a legal lens becomes ambiguous because there is no exact equivalent
of this term to be found in current EU data protection laws. Further-
more, when knowledge transfer is used in health data tasks, this practice
becomes even more complex, because health data qualifies as sensitive
data which attracts stricter rules regarding its processing. In this paper,
we examine this topic from both a legal and technical point of view.
Our research considers the use of repurposing machine learning models
and their application within the legal context of the secondary use of
personal data. Our legal analysis includes the General Data Protection
Regulation, Data Governance Act, and European Health Data Space. In
conclusion, we consider the advantages and disadvantages of the topic at
hand.