Medical Student University of Toronto Temerty Faculty of Medicine Toronto, Ontario, Canada
Background: Artificial intelligence (AI) encompasses a wide range of technologies that enables computers to perform tasks commonly associated with human intelligence. The rapid integration of AI in healthcare has caused changes in the way postgraduate medical education is conceived and delivered. Objective: To review the current applications of AI in pediatric postgraduate medical education programs and identify gaps in AI related scholarship for pediatric resident education. Design/Methods: A scoping review was conducted using a comprehensive literature search involving Ovid MEDLINE, OVID Embase, and ERIC conducted from January 1, 2000, until June 30, 2023. Inclusion criteria involved abstracts and articles that discussed AI in postgraduate pediatric education. Studies that addressed AI in undergraduate medical education or in specialities and subspecialities other than pediatrics were excluded. Information about the participants, study type, AI interventions used, and outcomes were abstracted from included studies and synthesized using a narrative approach. Results: The search yielded 983 unique publications and 16 met inclusion criteria. Six studies were conducted in the United States, three in China, three in France, two in Korea, one in Canada, and one in Germany. They investigated the use of AI in General Pediatrics, Pediatric Radiology, Pediatric Oncology, Pediatric Otolaryngology, and Pediatric Emergency Medicine programs. Two studies investigated machine learning models, two used Bayesian-based tools, 11 explored deep learning models, and one explored AI attitudes and training. AI as a clinical decision support tool in pediatric residency training was used in 13 studies including four showing improved diagnostic accuracy and nine showing better radiographic interpretation with AI-based training amongst residents. Two studies used AI models to assess competency of residents. One study assessed the experiences of AI amongst young pediatricians and found that only 5% of participants received AI training, while 87% considered implementation of such training to be necessary in postgraduate programs.
Conclusion(s): Overall, the studies included in this review demonstrated that AI tools improved residents’ workflow, diagnostic accuracy, and clinical confidence. However, studies lacked specific suggestions for program-wide AI curriculum delivery and raised concerns about the ethical and societal issues linked with the implementation of AI in residency training.
