Embedding ethical and legal principles in AI-driven clinical practice: two use cases in laboratory diagnostics

Abstract

This interdisciplinary paper explores the ethical-legal aspects of artificial intelligence (AI) in medicine. We first describe our ‘Embedded Ethics’ approach which entailed collaboration between an ethicist, legal scholar and AI specialist, and provide a brief overview of AI and the relevant European ethical-legal context. We then identify and analyse ethical and legal issues in clinical AI using two cases from the field of laboratory diagnostics: AI systems in newborn screening and for diagnosing Alzheimer's disease. These use cases reveal key concerns at each of the five phases of AI development and implementation. For every phase, we outline core ethical-legal principles and formulate guiding questions for developers. Concerns include fundamental questions about desirability and proportionality, AI-specific challenges like explainability and bias, and broader issues of shared decision-making and professional responsibility. We conclude with reflections on future challenges for clinical applications of AI systems and interdisciplinary research in this area.