Primer on large language models: an educational overview for intensivists.

The integration of artificial intelligence (AI) and machine learning-enabled medical technologies into clinical practice is expanding at an unprecedented pace. Among these, large language models (LLMs) represent a subset of machine learning designed to comprehend linguistic patterns, semantics, and contextual meaning by processing vast amounts of textual data. This educational primer aims to inform intensivists on the foundational concepts of LLMs and how to approach emerging literature in this area. In critical care, LLMs have the potential to enhance various aspects of patient management, from triage and clinical documentation to diagnostic support and prognostic assessment of patient deterioration. They have also demonstrated high appropriateness in addressing critical care-related clinical inquiries and are increasingly recognized for their role in post-ICU rehabilitation and as educational resources for patients' families and caregivers. Despite these promising applications, LLMs still have significant limitations, and integrating LLMs into clinical workflows presents inherent challenges, particularly concerning bias, reliability, and transparency. Given their emerging role as decision-support tools and potential collaborative partners in medicine, LLMs must adhere to rigorous validation and quality assurance standards. As the trajectory toward AI-driven healthcare continues, responsible and evidence-based integration of LLMs into critical care practice is imperative to optimize patient outcomes while ensuring ethical and equitable deployment.