Accountability Framework for Healthcare AI Systems: Towards Joint Accountability in Decision Making

Accountability means different things to different people: The paper found that "accountability is perceived and dealt with in many different ways, depending on the actor's expertise and domain of work," revealing confusion among healthcare professionals, AI developers, and regulators about what accountability actually means in practice.

Current regulations provide the "what" but not the "how": While regulatory bodies like the EU Commission offer high-level guidelines on what should be done, they create a "knowledge gap for actors" by failing to explain how healthcare organizations should actually implement accountability measures in their AI systems.

Healthcare AI accountability requires shared responsibility: Rather than placing accountability on a single actor, the framework advocates that "decision-making in healthcare AI holds shared dependencies, where accountability should be dealt with jointly and should foster collaborations" between clinicians, developers, regulators, and other stakeholders.

A three-tier structure helps organize accountability mechanisms: The paper provides a practical framework that "guides the actors of the healthcare AI system to categorise the mechanisms based on their conduct," making it easier for organizations to understand which accountability measures apply to their role.

Explainability plays a key role in accountability: The paper highlights "the role of explainability" as central to achieving accountability, suggesting that making AI decisions understandable to healthcare professionals is essential for responsible AI deployment in medical settings.