Abstract
Clinical simulation is key in medical education, but its traditional implementation is costly and complex. This study explored the use of ChatGPT (generative AI) to create interactive clinical scenarios and evaluated students' perceptions of this innovative tool. An exploratory, cross-sectional study was conducted with 32 fourth- and fifth-year medical students. A case of acute coronary syndrome was designed, where ChatGPT (GPT-4) acted as a virtual patient, dynamically responding to students' questions and decisions. After the simulation, participants completed a survey with 20 Likert-scale items and open-ended questions. The results showed high satisfaction: more than 90% considered the case clear and realistic, 94% positively evaluated the virtual patient's responses, and 88% perceived the difficulty as appropriate. Additionally, 97% found automated feedback useful. Qualitative comments highlighted the immersion and safety of the experience, allowing students to practice without fear of making mistakes. Some suggestions included adding greater clinical depth and adjusting the case difficulty. ChatGPT is a promising tool for clinical simulation, offering realistic and challenging scenarios that students perceive as beneficial for developing clinical reasoning and decision-making skills in a safe environment. Automated feedback was highly valued. However, this study is exploratory, and further research with larger samples is needed to validate its educational impact and address challenges such as content supervision and the ethical integration of AI in medical education.