Selcen Öncü, Fulya Torun, Hilal Hatice Ülkü
Effective physician-patient communication is essential for addressing HIV-related stigma and supporting patient-centred care. This study aimed to examine medical students' communication during an AI-supported patient interview in which ChatGPT-4o acted as a standardised patient living with HIV, with a focus on empathy and stigma-sensitive communication. A sequential explanatory mixed-methods study was conducted with third-year medical students at a public medical school in Türkiye. Of 277 students who completed a standardised, text-based AI-supported consultation, 260 completed the surveys and were included in the quantitative analysis; 30 participated in two focus-group discussions. Quantitative data were analysed using descriptive statistics, independent-samples t-tests and Pearson correlations, while qualitative data were thematically analysed using MAXQDA 2022. The highest mean item ratings concerned supportive and non-judgemental communication and appropriate responses to the patient's concerns. However, lower ratings were observed for providing accurate HIV-related information and using open-ended questions. A very weak positive correlation was found between students' self-assessment and educator assessment ratings (r = 0.136, p = 0.037). Qualitative findings highlighted empathy, ethical sensitivity and difficulty in formulating questions. AI-supported patient simulation may provide a complementary, preparatory opportunity for students to practise patient-centred and stigma-sensitive communication before patient contact.