The world of Artificial Intelligence (AI) is advancing rapidly. It seems only recently that healthcare was envisioning the possibilities of how AI and Machine Learning (ML) could support clinical practice. Physicians imagined being freed from the administrative burdens of medicine and gaining more time for meaningful human connection with patients. They also recognised opportunities to refine diagnosis and improve the delivery of treatment. This was particularly relevant for anaesthesiologists.
The concept of an anaesthetic machine capable of continuously monitoring a patient, without fatigue or distraction, by assessing depth of anaesthesia, haemodynamics, and neuromuscular function, while simultaneously titrating drugs appropriately, held enormous appeal. The development of closed-loop systems for anaesthetic agents and depth of anaesthesia came to represent the forefront of anaesthesia research and innovation.
One early attempt at semi-autonomous anaesthesia was the Sedasys pharmacological robot developed by Johnson & Johnson. It was designed to provide procedural sedation during colonoscopy without the presence of an anaesthetist in the room. Following years of development, Sedasys was ultimately withdrawn from the market after strong opposition from anaesthesiology professional societies. Although it demonstrated the promise of automation, limitations in adaptability and responsiveness contributed to its discontinuation.
AI has now developed to the point where it is on the verge of becoming integrated into perioperative care. Its applications extend well beyond automation alone. Anaesthesiology and critical care stand to benefit significantly from AI-supported systems. In the preoperative setting, AI can assist with risk assessment through the interpretation of physiological data, enabling improved planning and potentially better patient outcomes. During intraoperative care, AI-driven mechanical ventilation systems can adapt ventilation strategies to individual patient requirements, adjusting parameters in real time in response to physiological changes.
Automated sonography and predictive models may also assist clinicians by anticipating complications, optimising treatment pathways, and reducing the burden associated with time-critical care. In the postoperative setting, AI systems may help identify early clinical deterioration and provide timely warnings to support intervention.
While these developments appear highly promising, an important question remains: do these ambitions safely align with clinical realities?
In the symposium session at #EA26, “Algorithms in action: Artificial Intelligence (AI) at the frontlines of anaesthesia and intensive care”, speakers examined how AI is improving pre-anaesthesia evaluation through enhanced risk assessment and perioperative planning. The session explored recent advances in AI-driven mechanical ventilation, including its influence on respiratory mechanics, patient-tailored ventilation, and real-time adjustment of ventilation parameters. Speakers also discussed the integration of AI into automated sonography and predictive models, and the implications of these technologies for patient care.
Dr Rachele Simonte, an anaesthesiologist at Santa Maria della Misericordia Hospital in Perugia, Italy, discussed how AI and ML may enhance real-time ventilator management, including patient-specific adjustment of settings based on respiratory mechanics, closed-loop ventilation strategies, and integration with monitoring systems to support safer assisted ventilation.
Dr Regina Pikman Gavriely, an anaesthesiologist at Tel Aviv Medical Centre, Israel, examined how automated or AI-assisted sonography may provide real-time guidance for vascular access, perioperative respiratory assessment, and point-of-care ultrasound (POCUS) applications in anaesthesia and intensive care that are less dependent on operator expertise and experience.
Dr Carlos Ferrando Ortola, Head of the Surgical Intensive Care Unit, Department of Anaesthesiology and Critical Care at Hospital Clínic, Barcelona, Spain, focused on the use of AI and ML in predicting postoperative deterioration. He presented evidence demonstrating how ML models may be valuable in predicting early postoperative deterioration using preoperative and intraoperative data, alongside postoperative vital signs and laboratory findings, to estimate the likelihood of respiratory failure, re-intubation, or escalation of care to the ICU. He also discussed studies using ML to predict prolonged mechanical ventilation, ICU mortality, and outcomes in patients with AHRF/ARDS, all of which have significant relevance for anticipating postoperative deterioration.
Source: Euroanaesthesia 2026
Image Credit: iStock
