Page 9 - AIH-2-1

Artificial Intelligence in Health AI in higher medical education

A B

Figure 1. Geographical distribution of papers related to artificial intelligence (AI) in medical (med) education (ed) on the Web of Science. (A) Papers
including the terms AI+med. (B) Papers including the terms AI+med.

94 documents were taken into account. This investigation Overall, there is an underrepresentation of research in the
has two limitations. First, the study only takes into account developing world, despite the recognized importance of
the WoS database, which is the most restricted of its type AI+med+ed+ethics.
(although this ensures the integrity of the dataset), and A similar situation was found when a search for the
second, only publications written in English were included terms AI+med+radiology and AI+med+XR was conducted
in the systematic review, which may cause a potential (Figure 3A and B).
language bias.
Almost half of the occurrences of AI+med+radiology
The results of 328/488 resources may seem low. This
view is supported by Lee et al., who noted that AI is a were found in North America (49%), compared to only
12
7% in Europe. On the other hand, almost the opposite was
relatively new concept in medical education. More recently, found regarding AI+med+XR: North America (47%) and
as a result of an exhaustive search in four databases Europe (19%). This suggests that research and awareness
(PubMed, Embase, Scopus, and WoS) during the period of AI and radiology are more advanced in North America
2020–2024, Weidener and Fischer affirmed that there is than in Europe and that the opposite is true concerning
13
“a scarcity of literature on teaching AI ethics in medical AI and XR. Asian results were similar in both cases (25%
education, with most of the available literature being and 21%), but as a large developing territory, Africa was
recent and theoretical” (ibid., p. 399). The study shows substantially underrepresented (2% and 3%).
that the major studies (about 90%) in the field of AI ethics
were published in the years 2020–2024, which coincides 3. AI in medical education-some practical
with the dynamic development of AI. This is largely due applications
to the fact that currently solutions based on AI can be
implemented in practice, and there is a need to consider AI is increasingly seen as a significant resource for medical
all risks, both ethical and practical (technical). Since we education that will permeate all areas and become integral.
analyze the status of development and implementation AI is being applied in several different types of medical
of the general guidance on the ethics of AI in the field of fields, including technical support and distance learning,
medical education (with special emphasis on practical data analysis and interpretation, 3D modeling and remote
implications), in this study, we concentrate on the time virtual surgery, and text production by AI-powered text
frame in which the most dynamic development of the field generation engines (GenText) such as ChatGPT (Chat
of AI ethics occurs. In addition, the analysis highlighted a Generative Pre-trained Transformer). A study by Civaner
14
research gap in low-income countries. One of the reasons et al. showed that 80% of medical students perceive AI
may be the lack of access to the latest technologies, which as a technology supporting both the process of education
often involves significant costs. The lack of research in and health care, although the study also revealed concerns
this area also translates into a potentially low level of among the medical community about AI undermining
implementation of AI in practice. Indeed, it is evident that their skills and negatively impacting the patient-doctor
the results of AI+med+ed are a small proportion of those for relationship (50% and 40% of respondents, respectively).
AI+med, and that the results for AI+med+ed+ethics are an These concerns are not shared by biomedical physicians,
even smaller proportion of AI+med (Figures 2A and 2B). more than 80% of whom see AI as a support tool, not a risk

Volume 2 Issue 1 (2025) 3 doi: 10.36922/aih.3276