Resumo

ABRAHAM, R. The role of artificial intelligence (AI) in medical education: A pilot study on ChatGPT's impact on history-taking skills and objective structured clinical examination (OSCE) preparedness in undergraduate medical students. Afr. J. Health Prof. Educ. (Online) [online]. 2026, vol.18, n.1, pp.1-12. ISSN 2078-5127.  https://doi.org/10.7196/AJHPE.2026.v18i1.2711.

BACKGROUND. Effective communication, particularly history-taking, is essential for medical professionals to ensure accurate diagnoses and treatment plans. Traditional training methods, such as role-playing and standardised patients, can be resource intensive. Advances in artificial intelligence (AI) offer new solutions, with ChatGPT, an advanced AI language model, capable of simulating patient interactions, providing scalable and flexible practice opportunities that enhance clinical skills training. OBJECTIVES. This pilot study explores medical students' perceptions of using ChatGPT as a virtual patient for history-taking practice and evaluates its potential impact on objective structured clinical examination (OSCE) preparedness by comparing pre- and post-intervention OSCE scores. METHODS. Two hundred second-year medical students were invited to participate; 40 students completed the post-practice survey. These participants used ChatGPT (versions 3.5 and 4.0) to practice history-taking using prompts aligned with the Calgary-Cambridge framework. The survey evaluated usability, realism, skill improvement, feedback quality, and satisfaction. Separately, OSCE performance data for the entire class from the 2023 cohort (pre-ChatGPT) and 2024 cohort (post-ChatGPT) were compared using a two-sample t-test to assess differences in history-taking scores. Qualitative survey responses were thematically analysed. RESULTS. Among survey respondents, 85% found ChatGPT easy to use, 75% rated the scenarios as realistic, and 80% reported improvements in history-taking skills. Additionally, 78% felt better prepared for their OSCE. A statistically significant 4% increase in OSCE history-taking scores was observed between the 2023 cohort (mean 39/50, 78%) and the 2024 cohort (mean 41/50, 82%; p<0.05). Qualitative feedback highlighted ChatGPT's flexibility, ease of access, and immediate feedback, though students noted limitations in emotional nuance and occasional technical challenges. CONCLUSION. The improvement in OSCE history-taking scores at a class-wide level suggests a potential association between ChatGPT use and enhanced clinical performance. While only a subset of students completed the survey, their responses indicate that ChatGPT is a valuable supplement to traditional training, particularly for self-directed practice. These findings support the integration of AI into clinical education to strengthen communication skills and OSCE preparedness. Future research should include broader participation and long-term evaluations to fully determine the role of AI in medical education.

Palavras-chave : ChatGPT; virtual patient; artificial intelligence (AI); AI in medical education; history-taking skills; objective structured clinical examination (OSCE); clinical skills development; medical student perceptions; patient simulation; technology-enhanced learning; AI-driven learning tools; medical education innovation.

        · texto em Inglês     · Inglês ( pdf )