LETTER TO THE EDITOR

Exploring the Applications of ChatGPT in Family Medicine Education: Five Innovative Ways for Faculty Integration

Karim Hanna, MD

PRiMER. 2023;7:26.

Published: 8/2/2023 | DOI: 10.22454/PRiMER.2023.985351

To the Editor:

The integration of artificial intelligence (AI) tools in medical education holds promise for enhancing teaching and learning experiences. This letter focuses on the applications of ChatGPT (OpenAI) in family medicine medical education. ChatGPT uses natural language processing to generate humanlike responses and engage in interactive conversations.1 I write to encourage exploration of five innovative ways faculty can leverage ChatGPT to augment traditional teaching methods in family medicine education.

Medical educators face challenges in providing personalized learning experiences, enabling active student engagement, and addressing the wide spectrum of clinical scenarios encountered in family medicine.2,3 I have used the following methods in my experience with learners and believe that ChatGPT offers a unique opportunity to provide conversational agents that simulate patient encounters, facilitate clinical decision-making exercises, and foster critical thinking in learners. However, like any technology, it also has significant weaknesses that need discussion to mitigate.

  1. Virtual patient consultations. ChatGPT can be used to create realistic patient scenarios, allowing learners to practice clinical decision-making and communication skills in a simulated environment. This case-based learning with generated case studies or patient vignettes offers learners opportunities to apply their knowledge and reasoning abilities.
  2. Differential diagnosis exercises. Interactive exercises powered by ChatGPT can enable learners to input symptoms and receive potential diagnoses, facilitating the development of clinical reasoning skills. ChatGPT can be integrated into e-learning modules, enabling interactive educational content that provides instant feedback and responses to inquiries, and promoting self-directed learning.
  3. Clinical decision support. Faculty can use ChatGPT to provide real-time clinical decision support to students during patient encounters or case reviews, aiding in evidence-based decision-making.
  4. Exam preparation. ChatGPT can design practice questions and quizzes, allowing learners at all stages to test their knowledge and receive immediate feedback, thereby enhancing exam preparation. It also can assist in coaching through answering questions, as it is now being used across subject matters.
  5. Patient education materials. Learners can leverage ChatGPT to generate patient education materials, addressing common concerns about medical conditions and treatments, and providing clear explanations.

ChatGPT presents a range of exciting opportunities in medical education. It does so by facilitating case-based learning, supporting clinical decision-making discussions, and enhancing engagement through interactive and personalized learning experiences for learners and patients. Through its integration into educational contexts, faculty can augment teaching methodologies. I believe recognizing that ChatGPT should be used as a complementary tool, to reinforce traditional teaching methods, is important. AI systems rely on the data they are trained on; if the training data includes inaccurate information, a risk of propagating such misinformation exists.4,5 ChatGPT can generate outputs that seem plausible but are not entirely accurate or evidence-based. Content, including citations, can be fake, inappropriate, nonsensical, or irrelevant. For these reasons, human review is necessary to maintain quality. Establishing guidelines for the use of AI in medical education is crucial. By recognizing these weaknesses, we can harness the potential of AI in medical education.

References

  1. Ayers JW, Poliak A, Dredze M, et al. Comparing physician and artificial intelligence Chatbot responses to patient questions posted to a public social media forum. JAMA Intern Med. 2023;183(6):589-596. doi:10.1001/jamainternmed.2023.1838
  2. Frommeyer TC, Fursmidt RM, Gilbert MM, Bett ES. The desire of medical students to integrate artificial intelligence into medical education: an opinion article. Front Digit Health. 2022;4:831123. doi:10.3389/fdgth.2022.831123
  3. Shankar, PR. Artificial intelligence in health professions education. Arch Med Health Sciences. 2022;10(2): 256-261. doi:10.4103/amhs.amhs_234_22
  4. Tolsgaard MG, Pusic MV, Sebok-Syer SS, et al. The fundamentals of artificial intelligence in medical education research: AMEE guide no. 156. Med Teach. 2023;45(6):565-573. doi:10.1080/0142159X.2023.2180340
  5. Zawacki-Richter O, Marín VI, Bond M, Gouverneur F. Systematic review of research on artificial intelligence applications in higher education—where are the educators? Int J Educ Technol High Educ. 2019;16(1):39. doi:10.1186/s41239-019-0171-0

Lead Author

Karim Hanna, MD

Affiliations: Department of Family Medicine, University of South Florida, Morsani College of Medicine, Tampa, FL

Corresponding Author

Karim Hanna, MD

Fetching other articles...

Loading the comment form...

Submitting your comment...

By Keith  /  Posted 8/21/2023

This inspired me to register for the free ChatGPT3.5 and experiment a bit. A couple observations: - I entered in requests to give a summary on a couple recent topics I had done resident education on, including acute urinary retention. The result was amazingly quick and accurate, giving me a couple refinements in definitions that I would have liked to including in my 30 minute talk. - I entered in a request to give examples of how the ACGME FM milestones compared to the FM EPA's...and out came completely inaccurate correlations. It may have used the FM Milestones rather than the 2.0 FM Milestones that came out in 2019....but it I had not known a fair amount about the Milestones and EPAs....I would have been fooled by the confabulation. So - I am thinking the best application is to use Chat GPT on the back end, once one has already developed a presentation or paper, rather than relying on it to accurately generate content on the front end; and to not use Chat GPT at this point on content that you do not already have strong working knowledge of.

Downloads & Info

Share

Related Content

Tags

Searching for articles...