A Conceptual Framework for Critically Evaluating Integration of AI Diagnostic Support Systems into Clinical Practice

Abstract In our first section, we make the prior engagement between artificial intelligence (AI) and medicine explicit and show how pervasively AI (broadly conceived) has already influenced and been integrated into medicine and clinical care. There has been ethical reflection on these developments since the early 1960s, and many of the concerns articulated today (about impact on the physician-patient relation, explainability, bias and so on) have already been considered in some form long ago. In our second section, we briefly consider ML and what may be distinctive about diagnostic support systems utilizing ML. We also consider in very broad strokes some of the ethical challenges and technical tools offered within AI for addressing these challenges. Within this broader horizon, we consider in our third section the contributions made by the essays in the issue. We close with some summary thoughts on how reading these essays together sharpens our appreciation of ongoing ethical challenges for integrating diagnostic support systems into clinical practice.