The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index; 10 of the 14 sources its claims cite were checked for support.
Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology
Open accessAbstract
INTRODUCTION: This study aimed to evaluate the effectiveness of a generative artificial intelligence based simulated patient model in improving gynecological history-taking skills among medical students, compared with traditional clinical teaching methods. METHODS: A prospective randomized controlled trial was conducted involving 40 fourth-year medical students during their gynecology clerkship. Students were randomly assigned to either an AI group (n = 20) or a control group (n = 20). Both groups received identical theoretical instruction on history taking. The control group practiced interviews with instructor-led standardized patients, while the AI group conducted simulated interviews with a large language model configured to behave as a real patient using predefined gynecological cases. Outcomes were assessed using a gynecological history-taking content checklist, Objective Structured Clinical Examination (OSCE) Score, and the Mini-Clinical Evaluation Exercise (Mini-CEX). Statistical analyses were performed using t-tests, chi-square tests, or Mann-Whitney U tests. RESULTS: Baseline characteristics did not differ significantly between the two groups. Students in the AI group achieved a significantly higher history-taking checklist completion rate than those in the control group and missed fewer key history items. The AI group achieved significantly higher total OSCE scores compared to the control group. Mini-CEX results showed that the AI group performed significantly better in medical interviewing skills and organization/efficiency. No significant differences were observed in professionalism, clinical judgment, counseling skills, or overall clinical competence. CONCLUSIONS: The use of a generative AI-simulated patient significantly enhanced the information completeness and organizational quality of gynecological history taking among medical students. AI-based simulation appears to be an effective and scalable adjunct to traditional standardized patients, offering educational benefits for reinforcing systematic clinical inquiry. However, these findings are preliminary as a single-center study with a limited sample size. Further large-scale, multi-center research is required to confirm the generalizability of these results across diverse educational settings.
Overview
CoChat Summary
TL;DR
A prospective randomized controlled trial in 40 fourth-year medical students found that a generative AI–simulated patient improved gynecological history-taking completeness, OSCE scores, and some Mini-CEX interview/organization domains compared with instructor-led standardized patients. The authors conclude AI simulation is a promising scalable adjunct, but findings are preliminary due to single-center, small-sample, unblinded, short-term design.
Key findings
- AI group had higher history-taking checklist completion (93.07 ± 3.93% vs. 88.98 ± 4.05%, p = 0.0025, Cohen’s d = 1.03) and missed fewer key items (3.05 ± 1.73 vs. 4.85 ± 1.78, p = 0.0025, Cohen’s d = 1.03).
- AI group achieved higher total OSCE scores (89.43 ± 3.76 vs. 85.38 ± 3.62, p = 0.0010, Cohen’s d = 1.10).
- Mini-CEX showed better medical interviewing skills (7.70 ± 0.66 vs. 7.25 ± 0.64, p = 0.0342, Cohen’s d = 0.69) and organization/efficiency (7.60 ± 0.50 vs. 7.10 ± 0.64, p = 0.0092, Cohen’s d = 0.88).
- No significant differences were found in professionalism (p = 0.3572), clinical judgment (p = 0.6391), counseling skills (p = 0.6041), or overall clinical competence (p = 0.7739).
Methods
- Prospective randomized controlled trial with 40 fourth-year medical students during gynecology clerkship at Northern Jiangsu People’s Hospital Affiliated to Yangzhou University; 20 assigned to AI group and 20 to control group. Both groups received identical theoretical instruction.
- Control group practiced with an instructor-led standardized patient using calibrated illness scripts; AI group conducted text-based simulated interviews with ChatGPT-5.2 configured as a patient using preloaded gynecological case scripts, with temperature 0.7 and top-p 0.9.
- Outcomes were a 44-item gynecological history-taking content checklist, OSCE score, and Mini-CEX (9-point scale), graded by two independent faculty examiners; inter-rater reliability was Cohen’s κ = 0.83. Analyses used t-tests, chi-square tests, or Mann-Whitney U tests, with Cohen’s d and p < 0.05 in STATA 15.0.
Limitations
- Stated: small sample from a single institution limits generalizability; the study may have been underpowered to detect differences in Mini-CEX humanistic and cognitive domains, risking Type II error.
- Stated: examiners were not blinded to group assignment because the interventions differed in modality, which may introduce observer bias; human standardized patient and AI conditions also differed in standardization, and assessment was short-term without evaluation of skill retention or transfer to real clinical settings.
- Evident from the text: the AI interface was text-based and lacked non-verbal cues such as eye contact, grimacing, or body posture, which the authors note may explain the lack of superiority in empathy-related outcomes.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
5 Valid · 0 Not valid · 3 Not checked · 30 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Large language models (LLMs), such as GPT systems, can generate dynamic, contextually appropriate, and conversationally responsive simulated patients capable of mimicking a wide range of clinical scenarios [4, 5].”
§ Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology > Introduction
What the sources say
- Not checked — no full text of the sourcecites [4] Cook DA. Creating virtual patients using large language models: scalable, global, and low cost. Med Teach. 2025;47(20240711):40–2. https://doi.org/10.1080/0142159x.2024.2376879. source
- Partly supportedcites [5] Sardesai N, Russo P, Martin J, et al. Utilizing generative conversational artificial intelligence to create simulated patient encounters: a pilot study for anaesthesia training. Postgrad Med J. 2024;100:237–41. https://doi.org/10.1093/postmj/qgad137. sourceExperimental — accuracy not yet measured
The Methods describe an interactive simulated patient with a custom clinical backstory, and the interface description says she follows up based on trainees’ responses. The Randomness section reports varied answers across encounters. However, the pilot created one patient for an anaesthesia-training scenario; it does not demonstrate simulated patients across a wide range of clinical scenarios, and the Realism section notes limitations in how closely responses resemble real patient interactions. (LLM, conf=0.86)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Students in the AI group achieved significantly higher scores in medical interviewing skills compared with the control group (7.70 ± 0.66 vs. 7.25 ± 0.64, p = 0.0342, Cohen’s d = 0.69).”
- Also stated in: § Assessment of clinical competencies by Mini-CEX · plain
“| Medical Interviewing Skills | 7.7 ± 0.66 | 7.25 ± 0.64 | 0.0342* | 0.69 |”
§ Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology > Results > Assessment of clinical competencies by Mini-CEX
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Assessment of clinical competencies by Mini-CEX · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The mean number of missed key items was 3.05 ± 1.73 in the AI group, whereas the control group missed an average of 4.85 ± 1.78 items (p = 0.0025, Cohen’s d = 1.03).”
- Also stated in: § Performance on history-taking content checklist · plain
“| Missed key items (mean number) | 3.05 ± 1.73 | 4.85 ± 1.78 | 0.0025** | 1.03 |”
§ Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology > Results > Performance on history-taking content checklist
plainstates a causen = 20 ·- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Performance on history-taking content checklist · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The AI group achieved significantly higher total OSCE scores compared to the control group (89.43 ± 3.76 vs. 85.38 ± 3.62, p = 0.0010, Cohen’s d = 1.10).”
- Also stated in: § Performance on OSCE scores · plain
“| Total OSCE Score | 89.43 ± 3.76 | 85.38 ± 3.62 | 0.0010** | 1.10 |”
§ Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology > Results > Performance on OSCE scores
plainn = 20- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Performance on OSCE scores · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The AI group demonstrated a significantly higher checklist completion rate compared to the control group (93.07 ± 3.93 vs. 88.98 ± 4.05, p = 0.0025, Cohen’s d = 1.03).”
- Also stated in: § Results · plain
“Students in the AI group achieved a significantly higher history-taking checklist completion rate than those in the control group and missed fewer key history items.”
- Also stated in: § Performance on history-taking content checklist · plain
“| Checklist completion rate (%) | 93.07 ± 3.93 | 88.98 ± 4.05 | 0.0025** | 1.03 |”
§ Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology > Results > Performance on history-taking content checklist
boostedstates a causen = 20 ·- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Results · plain
12 more claims
Create a free account to see all 38 claims and every source verdict.
- It may state two separate claims as one. 17
- Probably not a substantial claim. 8
Citations
Cites
Kung, T. H., Cheatham, M., Medenilla, A., Sillos, C., De Leon, L., Elepaño, C., Madriaga, M., Aggabao, R., Diaz-Candido, G., Maningo, J., et al. (2023) · PLOS Digital Health · doi:10.1371/journal.pdig.0000198 · Cited by 4,115 · cited 1×
“Recent studies have highlighted the potential of LLMs to support clinical reasoning, generate patient narratives, and even demonstrate performance approaching passing thresholds on medical licensing examinations [6, 7].”
Ericsson, K. A. (2004) · Academic Medicine · doi:10.1097/00001888-200410001-00022 · Cited by 2,988 · cited 1×
“Besides, unlike human role-playing, where fatigue or variability may occur, the AI system offers repeated, high-fidelity practice opportunities [9, 11, 12], allowing students to master the check-list elements of history taking through deliberate practice [14].”
Research in clinical reasoning: past history and current trends
Norman, G. (2005) · Medical Education · doi:10.1111/j.1365-2929.2005.02127.x · Cited by 1,018 · cited 1×
“Clinical reasoning is a higher-order skill that typically develops longitudinally with clinical experience and case exposure, rather than through single simulation intervention [15].”
Kononowicz, A. A., Woodham, L. A., Edelbring, S., Stathakarou, N., Davies, D., Saxena, N., Tudor Car, L., Carlstedt-Duke, J., Car, J., & Zary, N. (2019) · Journal of Medical Internet Research · doi:10.2196/14676 · Cited by 600 · cited 1×
“This finding aligns with previous studies demonstrating that virtual patients and simulation-based education enhance learners’ ability to organize clinical information and reduce omission errors during data gathering [13].”
The use of simulated patients in medical education: AMEE Guide No 42
Cleland, J. A., Abe, K., & Rethans, J.-J. (2009) · Medical Teacher · doi:10.1080/01421590903002821 · Cited by 509 · cited 1×
“Moreover, standardized patient programs, although effective, can be resource-intensive and may not provide sufficient repetition for students to achieve fluency and confidence in gynecologic interviewing [3].”
Dornan, T., Littlewood, S., Margolis, S. A., Scherpbier, A., Spencer, J., & Ypinazar, V. (2006) · Medical Teacher · doi:10.1080/01421590500410971 · Cited by 501 · cited 2×
“Recognizing its critical role, medical courses world-wide prioritize history taking training, traditionally relying on communication-focused courses and clinical clerkships [2].”
Assessing empathy development in medical education: a systematic review
Sulzer, S. H., Feinstein, N. W., & Wendland, C. L. (2016) · Medical Education · doi:10.1111/medu.12806 · Cited by 294 · cited 1×
“This is likely because current text-based AI interfaces lack the non-verbal cues, such as eye contact, grimacing in pain, or body posture which are essential for triggering empathetic responses in real-world doctor-patient interactions [16].”
Shorey, S., Ang, E., Yap, J., Ng, E. D., Lau, S. T., & Chui, C. K. (2019) · Journal of Medical Internet Research · doi:10.2196/14658 · Cited by 238 · cited 1×
“Early educational research suggests that AI-generated virtual patients may improve learners’ engagement [8, 9], provide unlimited practice opportunities [10–12], and support consistent exposure to rare or sensitive clinical presentations.”
Holderried, F., Stegemann-Philipps, C., Herrmann-Werner, A., Festl-Wietek, T., Holderried, M., Eickhoff, C., & Mahling, M. (2024) · JMIR Medical Education · doi:10.2196/59213 · Cited by 152 · cited 1×
“Recent studies have highlighted the potential of LLMs to support clinical reasoning, generate patient narratives, and even demonstrate performance approaching passing thresholds on medical licensing examinations [6, 7].”
Creating virtual patients using large language models: scalable, global, and low cost
Cook, D. A. (2025) · Medical Teacher · doi:10.1080/0142159x.2024.2376879 · Cited by 73 · cited 1×
“Large language models (LLMs), such as GPT systems, can generate dynamic, contextually appropriate, and conversationally responsive simulated patients capable of mimicking a wide range of clinical scenarios [4, 5].”
Cited by
马, 基. (2026) · Advances in Education · doi:10.12677/ae.2026.1681811
Generative artificial intelligence and the transformation of medical education: a scoping review
Zhao, J., Yang, K., Gao, H., & Jiang, G. (2026) · Frontiers in Education · doi:10.3389/feduc.2026.1907589 · Cited by 1
Metadata
- Authors
- Gao, L., Zhang, L., Zhang, Y., Wu, M., Xu, J., & Lu, D.
- Affiliations
- Northern Jiangsu People's Hospital (CN)
- Corresponding author
- Gao, Lingling
- Venue
- BMC Medical Education
- Published
- 2026-05-04
- Volume
- 26
- Issue
- 1
- ISSN
- 1472-6920 (ISSN-L)
- PMC
- PMC13285479
- OpenAlex
- W7160051353
- Semantic Scholar
- b3392efd9a0b1128c42c59768ab969aa18124296
- Cited by
- 4
- References
- 16
- Citations by year
- 2026: 4
- FWCI
- 5.20
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by-nc-nd
- creativecommons.org/licenses/by-nc-nd/4.0 (tdm)
- creativecommons.org/licenses/by-nc-nd/4.0 (vor, 49 d)
- Locations
- BMC Medical Education (publishedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- Funders
- Research Project on Medical Teaching Reform at Yangzhou University.
- Topics
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Social Sciences › Education › Simulation-Based Education in Healthcare
- Social Sciences › Education › Innovations in Medical Education
- Keywords
- generative AI, standardized patients, medical interviewing skills, large language models, medical students, clinical rotations, OSCE, Mini-CEX, clinical competence
- MeSH terms
- Generative Artificial Intelligence, Adult, Artificial Intelligence, Clinical Competence, Education, Medical, Undergraduate, Educational Measurement, Female, Gynecology, Humans, Medical History Taking, Prospective Studies, Students, Medical, Patient Simulation
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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology. https://app.cochat.ai/papers/10.1186/s12909-026-09219-2
- The paper
- Gao, L., Zhang, L., Zhang, Y., Wu, M., Xu, J., & Lu, D. (2026). Utilizing generative artificial intelligence to create simulated patient for history taking in gynecology. BMC Medical Education. https://doi.org/10.1186/s12909-026-09219-2
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.