Artificial intelligence assisted telemedicine, clinical decision support for anesthesia and critical care in intensive care units: a scoping review

BACKGROUND: Artificial intelligence (AI) has been increasingly used in care delivery in intensive care units (ICUs) and anesthesia-critical care practice through telemedicine, tele-ICU systems, and remote patient monitoring, and is expected to support real-time clinical decision-making. METHODS: This scoping review followed PRISMA-ScR guidelines to map the existing evidence of AI in critical care and anesthesia-related ICU environments for telemedicine, telemonitoring, and clinical decision support systems. PubMed, Scopus, and Google Scholar were used to search for relevant literature, including the use of AI, telemedicine, predictive analytics, remote monitoring, and anesthesia-informed clinical decision support in critical care. RESULTS: The literature reviewed primarily focused on the non-generative AI solutions, such as machine learning, deep learning-based monitoring, and AI clinical decision support systems. Such systems can facilitate remote continuous monitoring, early detection of clinical deterioration, and clinical decision-making in the ICU perioperative anesthesia-critical care settings. The results were grouped into the following categories: tele-ICU implementation, predictive analytics, tele-monitoring, and AI-guided clinical decision support. The reported benefits included better monitoring, improved workflow, enhanced anesthesia and critical care decision-making, and greater access to specialist care, but there was substantial variation in the evidence of consistent improvement in patient-centered outcomes, with most of it being observational. Data quality, interoperability, model transparency, ethical issues, and lack of prospective clinical validation were the key difficulties encountered. CONCLUSION: AI-enabled telemedicine remains a nascent healthcare space in the ICU and anesthesia-critical care continuum, and further standardization, validation, and prospective clinical testing are needed to ensure its safe and scalable integration into clinical practice.