Artificial Intelligence in Emergency Medical Call-Taking and Dispatch Operations: A Scoping Review

Objective: This study aimed to systematically review worldwide research on artificial intelligence (AI) applications in emergency medical service (EMS) call-taking and dispatch operations and identify current research trends and research needs. Methods: This scoping review was reported in accordance with PRISMA-ScR. PubMed, Embase, Scopus, KoreaMed, IEEE Xplore, and the ACM Digital Library were searched for studies published from January 2021 to August 2026. Data were synthesized descriptively. Results: Eighteen studies from 13 countries and one multinational setting were included; eight (44.4%) were conducted in Europe. Patient Recognition and Classification was the predominant domain (11/18, 61.1%), followed by Dispatch Support and Operational Decision-Making (5/18, 27.8%) and Other Methodological Applications (2/18, 11.1%). Structured clinical or dispatch data were used in 11 studies, textual data in eight, and call audio in five; four applied multimodal integration. Deep learning (DL) was used as a primary model in 11 studies and classical machine learning (ML) in eight, including three using both; large language models (LLMs) were used in two. Eleven studies involved retrospective model development and internal validation, whereas only two were randomized controlled trials. None of the 11 model-development studies conducted independent external validation or calibration assessment, and the two randomized trials, one with high risk of bias and the other with some concerns, showed no consistent improvement across dispatch outcomes. Conclusions: AI applications span diverse clinical and operational tasks, with multimodal integration, speech technologies, and LLMs broadening their potential scope. Evidence remains dominated by retrospective model development and internal validation; prospective and real-world implementation studies are needed to establish safety and clinical utility before integration into EMS systems.