CoChat

Applications and Challenges of Implementing Artificial Intelligence in Medical Education: Integrative Review

Open access
Chan, Kai Siang, Zary, Nabil · 2019 · JMIR Medical Education

Published version of a preprint

The DOI matches its Crossref record; 59 of the 59 references it cites were found in a registry or an index; 8 of the 29 sources its claims cite were checked for support.

Abstract

BACKGROUND: Since the advent of artificial intelligence (AI) in 1955, the applications of AI have increased over the years within a rapidly changing digital landscape where public expectations are on the rise, fed by social media, industry leaders, and medical practitioners. However, there has been little interest in AI in medical education until the last two decades, with only a recent increase in the number of publications and citations in the field. To our knowledge, thus far, a limited number of articles have discussed or reviewed the current use of AI in medical education. OBJECTIVE: This study aims to review the current applications of AI in medical education as well as the challenges of implementing AI in medical education. METHODS: Medline (Ovid), EBSCOhost Education Resources Information Center (ERIC) and Education Source, and Web of Science were searched with explicit inclusion and exclusion criteria. Full text of the selected articles was analyzed using the Extension of Technology Acceptance Model and the Diffusions of Innovations theory. Data were subsequently pooled together and analyzed quantitatively. RESULTS: A total of 37 articles were identified. Three primary uses of AI in medical education were identified: learning support (n=32), assessment of students' learning (n=4), and curriculum review (n=1). The main reasons for use of AI are its ability to provide feedback and a guided learning pathway and to decrease costs. Subgroup analysis revealed that medical undergraduates are the primary target audience for AI use. In addition, 34 articles described the challenges of AI implementation in medical education; two main reasons were identified: difficulty in assessing the effectiveness of AI in medical education and technical challenges while developing AI applications. CONCLUSIONS: The primary use of AI in medical education was for learning support mainly due to its ability to provide individualized feedback. Little emphasis was placed on curriculum review and assessment of students' learning due to the lack of digitalization and sensitive nature of examinations, respectively. Big data manipulation also warrants the need to ensure data integrity. Methodological improvements are required to increase AI adoption by addressing the technical difficulties of creating an AI application and using novel methods to assess the effectiveness of AI. To better integrate AI into the medical profession, measures should be taken to introduce AI into the medical school curriculum for medical professionals to better understand AI algorithms and maximize its use.

Overview

CoChat Summary

TL;DR

This integrative review of 37 articles found that AI in medical education was used mainly for learning support, with assessment and curriculum review much less represented. The main implementation challenges were difficulty assessing AI effectiveness and technical difficulties in developing AI applications.

Key findings

  • Among 37 included articles, three primary uses were identified: learning support (n=32), assessment of students’ learning (n=4), and curriculum review (n=1).
  • For learning, the most commonly reported advantages were feedback for learning (21 articles), guided learning pathway/evaluation of learning process (18), decreased costs (8), no harm to patients (6), and less teacher supervision required (3).
  • Target audiences were medical undergraduates (n=25), postgraduates (n=14), and continuing medical education (n=8); 6 articles did not identify a specific target group. In the learning subgroup, medical undergraduates were the primary target audience (21/32).
  • Of 34 articles describing implementation challenges, the most common were difficulty assessing effectiveness (n=14), limited scalability (n=6), need for large sample size to develop the model (n=16), need for qualified content specialists (n=10), and physician-engineer communication gaps (n=4). One cited study reported a 22% posttraining improvement in diagnostic accuracy with an AI-driven simulator versus a nonstatistical 8% improvement with multimedia-based expert-led training.

Methods

  • Integrative review of peer-reviewed literature on AI in medical education, searching Medline (Ovid; 1954 to March 2019), EBSCOhost ERIC and Education Source (1983 to March 2019), and Web of Science (1986 to March 2019).
  • Inclusion focused on AI in doctors’ professional development across undergraduate, postgraduate, specialty training, and continuing medical education; exclusions included non-medical education, technology without AI, reviews/letters/commentaries, non-English full texts, and pre-1954 articles.
  • The search yielded 679 articles; after removing 185 duplicates, 494 were screened, 416 excluded, 78 assessed for eligibility, and 37 found relevant. Data were extracted and analyzed using the Extension of Technology Acceptance Model and Diffusion of Innovations theory, with quantitative pooling in SPSS where relevant.

Limitations

  • The authors state it may be too early to make an overarching statement about AI’s benefits in medical education because practices vary across subspecialties.
  • The authors note the high proportion of articles on AI as learning support compared with curriculum development/review and summative assessment, and state that conclusions about curriculum and assessment may be inconsequential due to the low number of studies reviewed.
  • Evident from the text: the search ended in March 2019, so later publications are not captured; the review also does not describe a formal quality appraisal or risk-of-bias assessment of included studies.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.2196/13930

Cites

59 of 59 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Total publications and sum of times cited by year in the last two decades. Retrieved from Web of Science for artificial intelligence in medical education, dated April 1, 2019.
Figure 1 Total publications and sum of times cited by year in the last two decades. Retrieved from Web of Science for artificial intelligence in medical education, dated April 1, 2019.CC BY · PubMed Central
Figure 2 Search strategy for literature on the use of artificial intelligence in medical education in undergraduate, postgraduate, and specialty training in medicine and beyond (continuing medical education). ERIC: Education Resources Information Center.
Figure 2 Search strategy for literature on the use of artificial intelligence in medical education in undergraduate, postgraduate, and specialty training in medicine and beyond (continuing medical education). ERIC: Education Resources Information Center.CC BY · PubMed Central
Figure 3 Subgroup analysis showing the number of articles in each focus group for the target audiences.
Figure 3 Subgroup analysis showing the number of articles in each focus group for the target audiences.CC BY · PubMed Central
Figure 4 Hierarchical presentation of the challenges of implementation of artificial intelligence (AI) in medical education. The upper blue rectangle shows the proportion of articles in each challenge category in the technical aspects of AI. The lower red rectangle shows the proportion of articles for challenges relating to perceived usefulness (in red) and perceived ease of use (in light red).
Figure 4 Hierarchical presentation of the challenges of implementation of artificial intelligence (AI) in medical education. The upper blue rectangle shows the proportion of articles in each challenge category in the technical aspects of AI. The lower red rectangle shows the proportion of articles for challenges relating to perceived usefulness (in red) and perceived ease of use (in light red).CC BY · PubMed Central

Verification

Claims the paper makes

7 Valid · 3 Not valid · 18 Not checked · 65 Own results and interpretations

57 more claims · 1 with their cited sources checked

Create a free account to see all 93 claims and to verify the 3 references still open.

See all claims — free account
  • Probably not a substantial claim. 18
  • It may state two separate claims as one. 20

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

8 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 59 references

Research works printed but not linked: 14

  1. McCarthyJMinskyMLRochesterNShannonCEAI magazine200612152019-06-07A proposal for the dartmouth summer research project on artificial intelligence, August 31, 1955 https://aaai.org/ojs/index.php/aimagazine/index
  2. Russell, S, 2018, Artificial intelligence: a modern approach, global edition
  3. Papa, F, 1992, Reports
  4. Billinghurst, M, 1996, Stud Health Technol Inform, 29, 590
  5. Frize, M, 2000, Clin Invest Med, 23, 266
  6. Voss, G, 2000, Stud Health Technol Inform, 70, 359
  7. StasiuRKDe BrittoJDa Silva DiasJScalabrinETeaching of electrocardiogram interpretation guided by a tutorial expertProceedings 14th IEEE Symposium on Computer-Based Medical Systems20010714th IEEE Symposium on Computer-Based Medical SystemsJuly 26-27, 2001Bethesda, MD48792
  8. Vicari, R, 2008, International Journal of Artificial Intelligence in Education, 18, A
  9. Kazi, H, 2009, International Journal of Artificial Intelligence in Education, 19, 309
  10. Chen, CK, 2010, IR Applications
  11. LemmonCSiuMLCVincentHVHamiltonJThe Importance of Humans in Simulation: Allowing the Lure of Technology to Drive DevelopmentProceedings of the European Conference on Games Based Learning2011105th European Conference on Games Based LearningOctober 20-21, 2011Athens, Greece
  12. Kutafina, E, 2015, Stud Health Technol Inform, 211, 286
  13. Khumrin, P, 2017, Stud Health Technol Inform, 245, 447
  14. Shin, J, 1993, CMAJ, 148, 969

Cited by

Items 201–300 of 746

Metadata

Authors
Chan, K. S., & Zary, N.
Affiliations
Nanyang Technological University (SG)
Corresponding author
Zary, Nabil
Venue
JMIR Medical Education
Published
2019-06-15
Volume
5
Issue
1
Pages
e13930
ISSN
2369-3762 (ISSN-L)
OpenAlex
W2939970150
Cited by
877
References
58
Citations by year
2026: 228 · 2025: 248 · 2024: 181 · 2023: 101 · 2022: 52 · 2021: 39 · 2020: 24 · 2019: 3
FWCI
1.57
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR Medical Education (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DR-NTU (Nanyang Technological University) (submittedVersion, cc-by)
Funders
Lee Kong Chian School of Medicine, Nanyang Technological University
Topics
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Social Sciences › Education › Artificial Intelligence in Education
  • Social Sciences › Education › Innovations in Medical Education
Keywords
artificial intelligence, medical education, AI implementation challenges, learning support, automated feedback, student assessment, curriculum review, medical undergraduates, technology acceptance, diffusion of innovations, artificial intelligence adoption, data integrity, big data

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-15, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-15 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.