Agentic education · Learning science

Actionable companion · Feedback & scaffolding

Learning with agency: ChatGPT Interactive Medical Simulations for Early Clinical Education: Case Study

AI-assisted editorial implementation guide, separate from the original research. Proposed actions are not claims or recommendations attributed to the source authors. No independent peer review of this guide is asserted.

Read original paperSearch paper agentCC BY 4.0

Evidence and context

Title and abstract screening; stage labels describe reported settings, not verified numerical ages. Design labels require full-text appraisal. Reported stage(s): Higher education. Health-professions learning must be distinguished from K–12 instruction.

Original authors’ abstract, reproduced under the linked source license:

Abstract Background The transition to clinical clerkships can be difficult for medical students, as it requires the synthesis and application of preclinical information into diagnostic and therapeutic decisions. ChatGPT—a generative language model with many medical applications due to its creativity, memory, and accuracy—can help students in this transition. Objective This paper models ChatGPT 3.5’s ability to perform interactive clinical simulations and shows this tool’s benefit to medical education. Methods Simulation starting prompts were refined using ChatGPT 3.5 in Google Chrome. Starting prompts were selected based on assessment format, stepwise progression of simulation events and questions, free-response question type, responsiveness to user inputs, postscenario feedback, and medical accuracy of the feedback. The chosen scenarios were advanced cardiac life support and medical intensive care (for sepsis and pneumonia). Results Two starting prompts were chosen. Prompt 1 was developed through 3 test simulations and used successfully in 2 simulations. Prompt 2 was developed through 10 additional test simulations and used successfully in 1 simulation. Conclusions ChatGPT is capable of creating simulations for early clinical education. These simulations let students practice novel parts of the clinical curriculum, such as forming independent diagnostic and therapeutic impressions over an entire patient encounter. Furthermore, the simulations can adapt to user inputs in a way that replicates real life more accurately than premade question bank clinical vignettes. Finally, ChatGPT can create potentially unlimited free simulations with specific feedback, which increases access for medical students with lower socioeconomic status and underresourced medical schools. However, no tool is perfect, and ChatGPT is no exception; there are concerns about simulation accuracy and replicability that need to be addressed to further optimize ChatGPT’s performance as an educational resource. Keywords: ChatGPT, medical school simulations, preclinical curriculum, artificial intelligence, AI, AI in medical education, medical education, simulation, generative, curriculum, clinical education, simulations

What kind of evidence is this?

Supporting AI learning science · Design requires full-text appraisal

Stage: Higher education. Screening labels describe reported context and study-design language. They are not a risk-of-bias rating or proof of effectiveness for an age band.

Study methods · source passage

To start, we selected two simulation categories with which to run prompts in ChatGPT. We chose advanced cardiac life support (ACLS) and 2 medical intensive care unit (ICU) scenarios—pneumonia and sepsis; ACLS was chosen because of the ACLS algorithm’s simplicity, while the medical ICU category was chosen for its potential complexity. Both simulations were inspired by Irvine School of Medicine’s “Clinical Foundations I” simulation curriculum at the University of California [11].

Limitations and uncertainty · source passage

ChatGPT has much to offer medical education. However, it also has flaws that complicate its potential implementation. Over the course of testing different opening simulation instructions, ChatGPT struggled to replicate simulation parameters. For example, ChatGPT occasionally provided feedback on the appropriateness of a decision and then progressed the simulation as if the correct decision had been made or asked multiple-choice questions instead of asking for free-response inputs. Small changes in punctuation and diction also seemed to have an effect, as did starting a new chat window within ChatGPT. More research on ChatGPT’s ability to be replicated and standardized is imperative if these simulations are to become a reliable tool.

Design the system before the pilot

Use this study to inform a learning mechanism or implementation question. It is not direct evidence that autonomous agents improve learning. Keep the proposed agent behavior separate from what the source actually tested.

  • Give feedback on the learner’s reasoning, then fade hints. Record whether the learner can complete the next task without that scaffold.
  • Use common rubrics, blind scoring where feasible and a fresh unaided reasoning task. Document permissible assistance and avoid treating AI-detector scores as proof of misconduct.

A 90-day district pilot

  1. Days 1–15 · Appraise. Have the curriculum lead and a teacher read the methods and limitations. Record the study design, sample, comparator, outcomes and evidence gaps. Define a local question that the study can help inform.
  2. Days 16–30 · Design. Choose one disciplinary task. Require a first attempt, a documented AI consultation, a critique of the output and an independent explanation. Align the comparison group and scoring rubric with the same objective. Name an accountable owner and set the baseline, success threshold and budget before implementation.
  3. Days 31–60 · Test. Start with a small voluntary cohort, preserve an accessible non-AI route and require teacher review. Use synthetic or de-identified examples; approve data handling and tool terms through district processes.
  4. Days 61–90 · Decide. Blind-score reasoning and accuracy on a new task without AI. Distinguish model examination scores, student perceptions and actual learner gains; only the last directly tests educational benefit. Compare with baseline, document uncertainty and stop or revise if learning worsens, access is unequal or critical errors persist.

The board decision brief

Report the learning problem, evidence quality, participating population, total cost including review time, measured benefit, errors, access differences and remaining uncertainties. Make a dated recommendation to stop, revise or expand. A positive perception survey alone does not establish learning improvement.

Attribution and reuse

Riley Scherr; Faris F Halaseh; Aidin Spina; Saman Andalib; Ronald Rivera. ChatGPT Interactive Medical Simulations for Early Clinical Education: Case Study. JMIR Medical Education (2023-11-10). https://doi.org/10.2196/49877

Companion prepared 2026-10-30. Original authorship and DOI are preserved. The abstract is an attributed source extract; the implementation plan is new AI-assisted editorial material, offered under the same CC BY 4.0 license. This guide has not undergone independent peer review.