Artificial Intelligence in cardiopulmonary resuscitation training – A scoping review

Abstract

Objectives This scoping review aimed to identify Artificial Intelligence methods used in cardiopulmonary resuscitation (CPR) training. Methods Members of the writing group ‘Education for Resuscitation’ of the European Resuscitation Council 2025 guidelines used the PICOST format for this scoping review, which included only published randomized and non-randomized studies. Medline, Embase, Cochrane, Education Resources Information Center, Web of Science, and PubMed were searched from inception to July 2025. Title and abstract screening, full-text review, and data extraction were performed by two researchers in pairs. PRISMA reporting standards were followed. The review was registered at PROSPERO. Because the evidence was insufficient for a systematic review, we changed our initial plan and performed a scoping review. Results The search identified 6977 citations. After removing 2521 duplicates, reviewing titles and abstracts yielded 43 articles for full-text review. Of these, 15 studies were included in the final analysis. Our findings reveal that Artificial Intelligence is being explored across key areas of CPR training, including its accuracy in detecting CPR quality parameters, providing real-time feedback, creating personalized training experiences, detecting and analyzing dialog segments during and after simulation, generating medical teaching illustrations, its capacity for interactive simulations, and answering laypersons’ medical questions. Conclusion Artificial Intelligence shows potential for transforming CPR training via enhancing real-time feedback, enabling personalized learning, improving dialog analysis, facilitating content creation, and serving as an information source. The current evidence is dominated by proof-of-concept studies. Future research needs to establish the efficacy of Artificial Intelligence-supported CPR training compared to traditional methods.