EP040 Artificial intelligence in anaesthesiology: a narrative review of applications, challenges, and future directions

<h3>Background and Aims</h3> Artificial Intelligence (AI) is reshaping anaesthesiology. This narrative review provides a comprehensive understanding of AI’s mechanisms, an extensive overview of current applications, and a critical examination of ethical, legal, and implementation challenges. It also highlights practical considerations for safe adoption and future integration opportunities. <h3>Methods</h3> A systematic literature search was conducted via PubMed on April 17, 2025, using the terms ‘Artificial Intelligence’[Mesh] AND ‘Anesthesia’[Mesh]. Full-text, English-language publications from the past five years were included. 126 out of 180 identified studies met eligibility after exclusion of retracted and inaccessible articles. Relevant references from academic lectures were also incorporated. <h3>Results</h3> AI is increasingly impacting anaesthesiology training and clinical practice. In training, AI-driven virtual environments promise improved resident engagement and skill acquisition. In preoperative assessment, machine learning (ML) models already surpass risk stratification tools, although clinical implementation remains limited due to interpretability and validation concerns. Intraoperatively, AI might support airway assessment, automated tracheal access, hemodynamic prediction and anaesthetic closed-loop systems. In regional anesthesia, AI potentially enhances ultrasound interpretation, image acquisition and procedural confidence. Postoperatively, ML can predict complications and enable remote monitoring. Integrating physiological, laboratory, and genomic data may provide the necessary infrastructure for developing human digital twins. (HDT). However, many applications lack prospective validation, standardized evaluation frameworks and present significant ethical, legal and logistical challenges that need urgent medical attention. <h3>Conclusions</h3> AI holds great promise for advancing anaesthetic care, but its clinical implementation remains limited by challenges in interpretability, data security, and a lack of prospective validation. Anaesthesiologists must critically assess these tools to ensure they improve patient safety and quality of care.