Abstract: Objective To investigate the application effect of combining Generative Artificial Intelligence (GenAI) with scenario-based simulation teaching for standardized residency training in the intensive care unit (ICU). Methods This controlled experimental study included 128 resident physicians who underwent standardized residency training in the Department of ICU at the Second Affiliated Hospital of Guangzhou University of Chinese Medicine (Guangdong Provincial Hospital of Chinese Medicine) from January to December 2024. Participants were grouped according to their rotation period: residents rotating from January to June 2024 were assigned to the control group (n = 63) and received traditional teaching, while those rotating from July to December 2024 were assigned to the experimental group (n = 65) and received GenAI with scenario-based simulation teaching. After the teaching intervention, training efficacy were evaluated by comparing end-of-rotation assessment scores, Mini-clinical evaluation exercise (Mini-CEX) scale scores, and teaching satisfaction questionnaire results between the two groups. Data were analyzed using the independent-samples t test, Chi-square tests, or Fisher′s exact test. Results The end-of-rotation assessment score of the experimental group (85.74±1.69) was higher than that of the control group (80.94±0.78), with a statistically significant difference (P<0.001). The Mini-CEX assessment showed that the scores of the experimental group in history taking (7.28±1.13), physical examination (7.54±0.87), clinical judgment (6.92±0.96), and overall clinical performance (7.65±0.82) were significantly higher than those of the control group (6.87±0.98, 7.10±1.13, 6.46±1.44, and 7.21±1.12), with all differences reaching statistical significance (all P<0.05). The teaching satisfaction survey showed that the scores of the experimental group in improving learning interest and efficiency (4.38±0.60), enhancing clinical work competence (4.43±0.61), and improving doctor-patient communication skills (4.32±0.73) were significantly higher than those of the control group (4.14±0.74, 4.14±0.74, and 4.03±0.82), with all differences statistically significant (all P<0.05). Conclusions GenAI-integrated scenario-based simulation teaching can effectively improve the training outcomes of standardized residency trainees in critical care medicine, and provide a reference for optimizing the teaching model of critical care residency training.

Key words: Intensive care, Standardized residency training, Generative artificial intelligence, Scenario simulation teaching