Artificial Intelligence in Parkinson’s Disease
Artificial Intelligence in Parkinson’s Disease

AIPD-DC7-UNamur

Exploring how the use of AI in medicine can be compliant with GDPR and AI Act

  • Host Institution: University of Namur
  • PhD Enrolment: University of Namur
  • Start Date: October 2025
  • Duration: 36 months
  • Official PhD Supervisor: Cécile de Terwange

Research Objectives

The use of AI in medicine raises questions about data protection on several levels, first in terms of data quality and individual consent. One objective of this project will be the reconciliation between the GDPR, including the duty of medical secrecy, and the use of AI in medicine. Europe aims to provide a framework for the use of AI and the initiative is entering the Trilog process. However, the proposal of the AI Act raises questions amongst actors in healthcare industry, such as mandatory quality standards for the datasets used by algorithms. Beyond this concern, the application of AI in medicine raises questions about obtaining consent from data subjects and the processing of data for secondary use, as outlined in the GDPR (article 5, b)) and the European Health Data Space (EHDS) proposal. Addressing these two issues (quality and consent), this project tackles the concept of anonymity as well as the use of synthetic data.

This project has three objectives:

  1. Explore the issues of quality and consent: Through rigorous analysis, the project aims to uncover speech signatures that are indicative of PD-related changes regardless of confounding effects. Bearing in mind that one way to fall outside the scope of the GDPR involves utilising anonymous data, this will also address the feasibility of achieving complete anonymity when handling health-related data.
  2. Explore the concepts of anonymity and synthetic data: Investigate the use of synthetic data in medicine as a means to mitigate the disadvantages associated with using anonymous data.
  3. Explore the liability in using AI: Through rigorous analysis of existing legislation, the project aims to propose adaptations to rules regarding liability in the use of AI.

The project will explore how to deal with:

  • The consent of a data subject when an algorithm uses his/her data.
  • The quality of the data due to the high risk of bias and the weakness of the sources.

A solution to counteract these two issues would consist of the use of synthetic data and the concept of virtual twin. Following Robert Riemann, “from a data protection by design approach, this technology could provide, upon a privacy assurance assessment, an added value for the privacy of individuals, whose personal data does not have to be disclosed” and “synthetic data might contribute to mitigate bias by using fair synthetic datasets to train artificial intelligence models. These datasets are manipulated to have a better representativeness of the world (to be less as it is, and more as society would like it to be). For instance, without gender-based or racial discrimination.” This project will promote the collaboration between data scientists and lawyers. DC7 will work in close collaboration with DCs4,5,6, because it will focus on speech biomarkers as a technically relevant use case. Furthermore, DC7 will collaborate with DC13 due to the common focus on synthetic data.

Expected Results

  • Innovative approach of the use of health-related data in medicine
  • Innovative approach of liability in an AI environment
     

Planned Secondment(s)

  • Host: ki:elements
    • Duration: 8 months
    • Purpose: Better understanding of data scientist perspective on data quality, anonymization and synthetization

This project is part of the "Trustworthiness" work package.