Artificial intelligence (AI) is entering clinical practice through decision-support systems, predictive tools, generative models, and clinical documentation solutions. Since February 2025, the European AI Act has required providers and deployers of AI systems to ensure that staff and other people operating or using such systems on their behalf have an adequate level of AI literacy. This creates a new educational requirement: physicians need to acquire minimum competencies to use, appraise, supervise, and reject AI outputs when appropriate. This state-of-the-art review examined AI competencies relevant to postgraduate medical training, focusing on literature published since 2022. The reviewed literature suggests thematic convergence around six core domains: foundational AI literacy, critical appraisal of AI tools, safe clinical application, ethics/law/governance, patient communication, and health data literacy. Other domains, such as institutional implementation, multidisciplinary collaboration, and AI leadership, appear as intermediate or advanced competencies. We propose a three-tier curricular organization: baseline competencies for all physicians; proficient competencies for physicians involved in local appraisal and implementation of AI systems; and advanced competencies for clinician-scientists, institutional leaders, and professionals with formal responsibilities in AI governance. In the Portuguese context, this structure may support the development of transversal training for residents and specialists, aligned with technological change, European regulatory requirements, and the need to preserve human clinical responsibility. This proposal should be understood as a conceptual basis for multidisciplinary consensus validation.
João Frutuoso, A. Maria, Helena Donato et al.· Acta Médica Portuguesa· 0 citations
The rapid expansion of artificial intelligence (AI) in the health sector has intensified the need for a robust regulatory framework capable of safeguarding patient safety, professional accountability, and public trust. We aimed to analyse the European Union Artificial Intelligence Act [AI Act, Regulation (EU) 2024/1689] and its implications for the healthcare sector in Portugal, presenting the institutional position of the Artificial Intelligence Committee of the Portuguese Medical Association (CPIA-OM). We conducted a narrative review based on official European legislation [AI Act, General Data Protection Regulation (GDPR), Medical Device Regulation (MDR) and In Vitro Diagnostic Medical Devices Regulation (IVDR)], institutional documents [including the 2025 White Paper on AI in Healthcare in Portugal, published by the Shared Services of the Ministry of Health (SPMS)], and peer-reviewed literature on artificial intelligence in medicine, ethics, and regulatory science. The AI Act introduces a risk-based classification of AI systems, ranging from prohibited "unacceptable risk" applications to minimal risk uses. Most medical AI applications, including diagnostic support, clinical monitoring, and therapeutic decision-making, are classified as "high risk" and are subject to stringent requirements regarding conformity assessment, technical documentation, human oversight, and data governance. The regulation interacts with existing frameworks such as the GDPR and MDR/IVDR, creating complex compliance obligations. Specific challenges include algorithmic bias, liability attribution, and the preservation of the physician-patient relationship. In Portugal, recent institutional initiatives within the National Health Service - including pilot projects in triage, frailty assessment, antibiotic stewardship, and medical imaging - illustrate opportunities for responsible AI integration, while academic reports highlight persistent concerns around transparency, trust, and professional substitution. The AI Act represents a landmark in European health regulation, balancing innovation with fundamental rights protection. For Portugal, its implementation requires coordinated action between the government, regulators, healthcare providers, and professional bodies. The CPIA-OM emphasises the need for continuous medical education in AI, the establishment of clinical oversight protocols, and the creation of observatories to monitor implementation and ensure ethical, safe, and patient-centred adoption.
João Frutuoso, Juan Pereira, M. Pereira et al.· Acta Médica Portuguesa· 0 citations
