Academic and professional background, key areas of expertise, and research interests

Obtained a PhD at KU Leuven on the research topic of wireless EEG sensor networks, focusing on energy-efficient deep learning models. Currently a postdoc at KU Leuven investigating the usage of AI in several biomedical applications.

What is your role within the AI-PROGNOSIS project, and how do your expertise and skills contribute to achieving its objectives?

My current objective within AI-PROGNOSIS is the development of models that predict secondary outcomes such as dyskinesia’s or hallucinations and how medication influences them. To this end I apply machine learning methods to Critical Path for Parkinson’s Integrated Parkinson’s Database.

What future opportunities do you see for your field in similar healthcare projects?

The methodology for predicting such secondary outcomes is generic and not limited to the application for Parkinson’s disease. As such there are ample opportunities for this line of research in different healthcare projects.

What do you see as the long-term potential of the AI-PROGNOSIS project in advancing AI-driven healthcare solutions?

In the long term, AI-PROGNOSIS can serve as a major example on how AI can influence the full process of diagnosis, prognosis and treatment, kickstarting similar efforts in other medical fields.

From a clinical perspective, how do you think the AI-PROGNOSIS tools will influence the diagnosis, prognosis, and treatment of Parkinson’s disease?

AI-PROGNOSIS tool can influence this whole process in multiple ways. Cheaper, more accessible biomarkers can allow for easier, far faster detection of upcoming Parkinson’s disease and allow for better treatment at an earlier stage. Personalization of proscribed medicines can minimize harmful side-effects and ensure no more medicine is taken than necessary.

How do you see digital tools, such as apps, wearables, or other technologies, transforming healthcare in projects like AI-PROGNOSIS? What specific benefits do you anticipate for patients or healthcare providers?

The most important benefit I see in this regard is streamlining of the entire medical process. Digital tools already analyzed by an automatic system will allow for earlier detection of at-risk patients and provide a lot of information to the doctor before the patient even sets foot in the room, allowing for healthcare providers to focus more of their efforts on the personal guidance of the patients, providing context and explanations to them.

How do you envision AI specifically supporting the AI-PROGNOSIS project, and what direct benefits do you anticipate from its integration?

AI will be necessary to work through the vast amount of data available for the objectives of the project. Large language models can be very helpful in standardizing the very differing formats of various datasets, allowing us to easier pool them in one huge, integrated dataset.

What do you think is the wider impact of artificial intelligence in healthcare, and how could projects like AI-PROGNOSIS help drive this change?

On one hand, artificial intelligence will help with faster, automated detection of previously known patterns for diagnosis. This also encourages the steps being taken to acquire the necessary data for them will become more of a standardized routine. On the other hand, artificial intelligence will also help in the discovery of new patterns and establish new diagnostic practices. Projects like AI-PROGNOSIS can establish important proof-of-concepts to convince steakholders in other medical fields of the advantages of this transformation.

How are user research and co-creation part of the AI-PROGNOSIS project, and what benefits do you expect from working together in these ways?

AI tools will ultimately never be used fully autonomously and will serve a supporting role for medical practitioners. It is thus vital that these tools meet their demands, can explain the outcomes they obtain to a satisfactory degree and convince the end-users of the advantages of adopting them. Close collaboration with medical practitioners will also allow for a much smoother development of these models, as they have a much better knowledge of the relevant features to be included in the model and can help discern whether what the model is doing makes sense or not.