Practicing pediatric interviews with parents through conversational AI-based virtual patients: an observational study in Spanish undergraduate medical education

To evaluate students' acceptance of training via artificial intelligence (AI)-based virtual patients (VPs) in pediatrics; to analyze the main factors that affect student satisfaction; and to compare student satisfaction with that reported in the literature.

Two observational studies were carried out. Study S1 analyzed students' interactions with the platform and study S2 analyzed their answers to a final questionnaire. All students enrolled in the "Pediatrics II and Pediatric Surgery" course were invited to participate during a continuous session on May 7, 2025. Study S1 analyzed whether the case/session order, the number of interactions, the time spent, and student gender were associated with student satisfaction (by fitting a linear mixed-effects model); and also performed a qualitative analysis of students' open-ended comments. Study S2 analyzed the influence of demographic data on students' feedback (by fitting proportional odds ordinal logistic regression models and linear models). All experimental data, code and results are available.

Concerning study S1, 70 students participated. Platform rating was high (mean=9.04, standard deviation=1.09) and was positively associated with the number of interactions with the VPs (P=0.038, β=0.022 [0.001-0.043]). Concerning study S2, 60 students participated. No statistically significant influence of demographics was found either for answers to individual questionnaire items or for grouped answers.

Student acceptance of training with AI-based VPs was supported by the satisfaction outcomes measured and by comparison with previous studies. However, further research is needed to confirm these findings.