SISTEME DE ASISTENŢĂ PENTRU LUAREA DECIZIILOR CLINICE BAZATE PE INTELIGENŢĂ ARTIFICIALĂ: CAPACITĂŢI ŞI RISCURI JURIDICE ÎN CONTEXTUL REGLEMENTĂRII UNIUNII EUROPENE
ARTIFICIAL INTELLIGENCE-BASED CLINICAL DECISION SUPPORT SYSTEMS: LEGAL CAPABILITIES AND RISKS IN THE CONTEXT OF EUROPEAN UNION REGULATION

Summary/Abstract: This article provides a conceptual analysis of artificial intelligence-based clinical decision support systems (AI-CDSS) as a class of digital tools that influence diagnostic and therapeutic choices while simultaneously shifting part of the clinical risk into the algorithmic domain. The relevance of the topic is determined by the rapid implementation of machine learning in medical information systems and by the increasing number of situations in which model-generated recommendations become a significant factor in clinical decision-making. The aim of the study is to develop an analytical framework for correlating the clinical benefits of AI-CDSS with their legal and ethical limits and for identifying the conditions of their legally safe and clinically justified use. The methodological basis combines doctrinal legal analysis, comparative legal analysis of European and international regulatory approaches to software as a medical device, including in the context of EU regulation (Reg. (EU) 2017/745, 2024/1689), and a risk-based interpretation of AI-related clinical studies with particular attention to external validity, decision errors, and model transferability. The discussion focuses on three interconnected domains: data quality and model drift; organizational integration and cognitive effects, including automation bias; and allocation of responsibility together with the traceability of decisions. The scientific novelty of the article lies in substantiating the thesis that the key limitation of AI-CDSS effectiveness is determined not only by algorithmic accuracy, but primarily by the absence of institutionalized procedures for validation, monitoring, traceability, and incident management. The practical significance of the study lies in proposing implementation principles aligned with clinical safety, patients’ rights, data protection requirements, and legal compliance.

• Details
• Contents