Research in Medical Education

Technological advancements over the past decade have led to a fundamental transformation in the education domain. Among these, large language models (LLMs) such as ChatGPT with their ability to comprehend and generate natural language have opened new horizons for interactive and personalized learning (1).
Medical education, by its very nature complex, interdisciplinary, and requiring continuous interaction is one of the most suitable fields to benefit from the capabilities of these models. However, the effective utilization of LLMs necessitates a newly emerging skill known as prompt engineering. Prompt engineering refers to the deliberate, purposeful, and structured design of inputs that can maximize the efficiency and accuracy of language model outputs (2). This skill is not only a technical competence but also a cognitive and pedagogical capability that requires deep understanding of learning objectives, domain-specific language, instructional design, and the functional capacities of language models (3).
In the context of medical education, prompt engineering can play a pivotal role in designing clinical scenarios, practicing communication skills, generating diagnostic tests, and enhancing critical thinking. For instance, a medical student may, by crafting a well-structured prompt, ask a language model to present a realistic scenario for managing a patient with chest pain and subsequently analyze each step of clinical decision-making. Prompt engineering is thus the purposeful, precise, and structured design of inputs to language models with the aim of eliciting output. This skill integrates domain knowledge, linguistic competence, and awareness of how Artificial Intelligence models process language. In medical education, prompt engineering must be carried out with scientific sensitivity, conceptual precision, and clinical awareness (4).
In this regard, prompts must not only utilize specialized medical terminology but also follow an instructional structure that facilitates conceptual understanding, practice of clinical decision-making skills, and the simulation of real-life medical situations. For example, a prompt such as, “A patient presents with signs of hyperthyroidism; how would you confirm the diagnosis and initiate treatment?” holds greater educational value compared to a general prompt like, “Explain hyperthyroidism,” as the former clearly establishes context, role, and objective. To craft an effective prompt in medical education, the following strategies are recommended:
1) Clearly define the educational objective: It must first be determined what the intended outcome of the interaction with the model is conceptual learning, scenario analysis, or decision-making practice. 2) Provide adequate contextualization: Language models tend to make errors when given insufficient context.
3) Specify the response format: In many cases, if the desired output should be in the form of a table, a list, or a 200-word explanation, this should be explicitly stated. 4) Test and revise: Prompts should be tested for effectiveness. If the model’s response does not align with expectations, prompt revision is necessary. 5) Issue clear instructions: Specific guidance should be provided regarding the expected response style or format (5).Designing prompts that yield precise, scientifically grounded, and practical outputs requires sustained practice, feedback, and continuous refinement. This process may be time-consuming and confusing for novice instructors or students (6).

Prompt engineering is a powerful tool that can serve as a pedagogical complement to both instructors and patients in medical education. Through intentional training in this skill, language models can be leveraged for personalized learning, clinical skills practice, and enhancement of medical education.
Educators must embrace innovative educational tools and use them to enhance teaching quality and student engagement. Artificial intelligence is not a fleeting trend; those who resist its adoption risk falling behind in the trajectory of future learning. The future of medical education demands the creative integration of such technologies with traditional and value-based approaches.