JMIR Publications

ABSTRACT

Background:

In the response to major infectious disease outbreaks, the emergency care capabilities of healthcare workers directly impact the effectiveness of prevention and control efforts [1]. However, training systems have long faced the challenge of a “disconnect between learning and application,” meaning that what is learned in training is difficult to effectively apply in real-world, high-pressure scenarios. This phenomenon is defined in transfer of training as a transfer barrier, referring to an individual’s inability to translate acquired knowledge, skills, and attitudes into professional practice [2]. The classic model proposed by Baldwin and Ford [3] indicates that transfer of training effectiveness is influenced by the interaction of three factors: trainee characteristics, training design, and the work environment. In medical emergency contexts, these factors take on unique characteristics: high-stress states can lead to procedural deviations, personal protective equipment limits the precision of movements, and there are significant contextual differences between training and real-world environments. Understanding emergency medical training can be approached from two dimensions: first, technical emergency training for sudden incidents; and second, systematic training aimed at building long-term emergency response capabilities. This paper focuses on the latter, specifically exploring how to systematically address barriers to the transfer of training in the development of long-term emergency response capabilities. In recent years, artificial intelligence (AI) technology has been gradually integrated into medical education, providing new tools to address the “disconnect between learning and application” [4]. However, its application has largely focused on optimizing training design, and its ability to address the individual characteristics and work environment factors outlined in the aforementioned model remains unclear. Given the fragmented nature of research in this field, the diversity of intervention measures, and the lack of uniform standards for transfer assessment, this study employs a scoping review methodology to systematically summarize the current landscape of AI interventions in the transfer of emergency first aid training, with the aim of providing a foundation for future targeted interventions.

Objective:

Objective:

Emergency first aid training has long been plagued by a “disconnect between learning and application”—a barrier to learning transfer. While artificial intelligence (AI) technology is gradually being integrated into medical education, its effectiveness in improving transfer of training remains unclear. This review aims to systematically summarize the current state of research on AI interventions for transfer of training in emergency first aid training.

Methods:

Methods:

A systematic search was conducted in PubMed, Web of Science, CNKI, VIP, and Wanfang databases, covering the period from the establishment of each database to April 2026. Search terms included Chinese and English subject headings and free-text keywords such as “emergency first aid” “transfer of training” and “artificial intelligence” . A total of 72 studies were included.

Results:

Results:

Core barriers to transfer in emergency first aid training include rapid skill decay, failure to recall skills under stress, and insufficient organizational support. The application of AI technology can be categorized into four types: high-fidelity VR/AR simulation, intelligent real-time feedback, AI tutors/adaptive learning, and learning behavior analysis. Current evidence reveals a pattern of “strong near-transfer but weak far-transfer”: AI can significantly enhance skill acquisition and near-transfer effects in simulated environments, but less than 10% of studies address far-transfer (i.e., real-world emergency response behavior), and longitudinal designs are lacking; Measurements of transfer face structural issues such as insufficient ecological validity, fragmented indicators, and the absence of a unified assessment framework.

Conclusions:

Conclusion: AI holds application potential for enhancing near transfer, but there is currently no high-quality evidence to support its ability to independently resolve the challenge of far transfer from the transfer of training to real-world practice. Future research must shift the paradigm from verifying technological feasibility to validating transfer effectiveness, conduct longitudinal tracking, and develop standardized transfer measurement tools.