Agentic systems in emergency medicine: a comprehensive review

Clinical decision-making in emergency medicine is often hindered by time pressure and cognitive overload. Agentic and AI-based systems offer support, but real-world adoption is limited by technical, ethical, and operational challenges. This review explores: the design of agentic systems for real-time decision-making in Eds; safety and ethical considerations in deploying autonomous agents; and their impact on workflow and patient outcomes. A structured review of 33 peer-reviewed articles (2005–2025) from Scopus was conducted, focusing on AI and agentic decision support in emergency medicine. Articles were thematically analyzed across system design, governance, and clinical impact. Agentic systems such as task-based, mobile, and cooperative agents support flexible, real-time decision-making. Effective design integrates shared and recognition-primed mental models. Key challenges include ensuring human oversight, transparency, and clinician trust. AI improves workflow efficiency, but evidence on patient outcomes is mixed and adoption remains constrained by ethical and usability concerns. Agentic systems show strong potential to enhance emergency care but require rigorous validation, ethical safeguards, and better alignment with clinical workflows. Future work should focus on standard evaluation metrics, bias mitigation, and building user trust.