On March 31, the UvA's RPA AI for Health Decision-making hosted a grand kick-off event at Science Park, attracting participants from diverse backgrounds, including law scholars, computer scientists, and industry researchers in Medical AI. The RPA's main goal is to bring together interdisciplinary research in Health AI, and the diverse attendance highlights the importance of interdisciplinary collaboration in the field.
Researchers from the RPA presented their work, including their joint project investigating the quality and reliability of reporting on publicly available medical datasets. The project engages all PhD students from the RPA and scrutinizes the contents of a dataset, as well as the documentation on data collection, structuring, labeling, preprocessing, and distribution.
In the morning session, Sijm Noteboom (Medicine) presented on structuring and storing data collected in the ICU in cloud environments to gain more insight into clinically relevant questions. Maria Galanty (Science) discussed predicting prolonged atrial fibrillation to assist medical personnel in decision-making. Both speakers highlighted various interdisciplinary challenges, including medical, ethical, and legal issues. Sanne van Velzen (Science) discussed her work on identifying patients at risk of heart failure from a single lead ECG.
In the afternoon, Dieuwertje Luitse (Media Studies) presented on critical research on AI system and application development in Medicine, highlighting the importance of qualitative humanities research as part of the interdisciplinary RPA. She also presented her research into platform-based AI competitions that structure power relations in AI-based medical imaging and her ethnographic research into the politics of machine-learning dataset development in the ICU. Saar Hoek (Law) presented on explainability in AI and informed consent, emphasizing the need for the development of an "explanation framework" through qualitative interviews to analyze what types of explanation are necessary when AI is implemented in a clinical setting.
The event also featured three internationally acknowledged keynote speakers discussing cutting-edge research developments and major challenges in Health AI from their respective research angles: cardiology, medical image analysis, and health law. Professor Dr. Folkert Asselbergs (Amsterdam UMC) emphasized the critical need for novel infrastructures and guidelines to facilitate the comparability and compatibility of data sources across hospitals and countries for Health AI research purposes. Dr. Andrew King (Kings College London) discussed research into algorithmic bias and methods for bias mitigation in machine-learning driven medical imaging. Professor Dr. Stefaan Callens (KU Leuven) closed the day by framing legal perspectives on Artificial Intelligence for Medical Decision-making directly in relation to the GDPR and the AI Act as proposed by the European Commission.
Overall, the event provided a forum for discussing the latest research developments and interdisciplinary challenges in Health AI, highlighting the need for cross-disciplinary collaboration, ethical considerations, and reliable data infrastructures. Presentations and discussions also shed light on the potential benefits of AI for decision-making in the medical field, as well as the potential risks associated with algorithmic bias, explainability, and data privacy.
We closed the event with drinks!
