Speakers
Overview
Looking into how AI can support decision-making in clinical and maternity care contexts, one talk introduced Bayesian network models used for clinical decision support, focusing on the IMRisk model for estimating lymph node metastasis risk in endometrial cancer, and discussed open issues related to the AI Act. A second talk presented ongoing research on shared decision-making in maternity care, examining how AI could help patients clarify values and navigate preference-sensitive decisions. A third talk shared interview findings from a study with fifteen women who recently gave birth, identifying themes of information overload, emotional barriers, and the need for tailored support across the maternity journey.
Key Points
Bayesian networks are white-box AI systems where every parameter is interpretable; nonetheless, we still need to consider how to handle subjective outcomes like quality of life in a relative scale.
The AI Act requires clinical decision support systems to be explainable, but what explainability means for the patients, for the clinicians, and for the regulators is still being constructed, together with the necessary laws.
Shared decision-making in maternity care is challenging because many decisions have no clearly superior option; patients often feel overwhelmed by the sheer amount of information from care providers, family, and the internet.
Women in the interview study valued in-person contact with their care provider above all, so AI should support, not replace, this relationship, and is best positioned before or after a consultation rather than during it.
The main concerns raised by participants included data privacy, the environmental cost of AI, and who ultimately owns and benefits from sensitive health data.
Open question: How do we design AI tools that genuinely center patient values, run in local and privacy-preserving environments, and stay under human supervision?
Next Steps
Possible collaborations: linking neurocognitive mechanisms and clinical outcomes with ethical decision frameworks, and developing decision-support tools that address both empirical and normative uncertainty in healthcare.
If you are interested in AI-supported clinical or maternity decision-making and would like to connect with the speakers or suggest a speaker for another session, contact us at centerfordecisionscience@ru.nl.