To address the prominent issues in graduate education in critical care medicine, such as insufficient teaching resources and the disconnection between theory and practice, this study explores an innovative teaching approach based on artificial intelligence (AI) to enhance graduate students' clinical reasoning and emergency response capabilities.

A research team was established at the Department of Critical Care Medicine, Tianjin Medical University General Hospital. Fifty real and typical clinical cases from 2019 to 2023, covering all core diseases in critical care medicine, were selected. After double-blind desensitization, these cases were transformed into teaching cases suitable for graduate training using AI and classified into three difficulty levels: elementary (basic cases), intermediate (multidisciplinary collaborative cases), and advanced (difficult/technologically innovative cases). A knowledge graph was constructed to enable dynamic case generation and personalized case recommendations, and an intelligent teaching platform was built, integrating functions such as pre-test-based stratified learning, virtual emergency scenario simulation, and mind mapping for comparing with expert decision-making. A three-stage teaching process of "pre-class preparation—in-class simulation—post-class reflection" was adopted. Thirty-three graduate students majoring in critical care medicine from the 2022–2024 cohorts of Tianjin Medical University were selected as the experimental group (using the intelligent platform + novel teaching model), and 34 graduate students of the same major from the 2021–2023 cohorts served as the control group (adopting the traditional teaching model). There were no statistically significant differences in baseline data between the two groups (P＞0.05), indicating comparability. The observation indicators included the scores of case analysis questions in the final examination, and the results of a questionnaire survey on students' satisfaction with the teaching model and evaluation of ability improvement.

The average score of the experimental group on case analysis questions was 75.94, which was significantly higher than that of the control group (65.64), with an improvement of 10.3 points. The questionnaire survey results showed that 81.8% (27/33) of the experimental group students believed the model facilitated the deepening of theoretical understanding and application, 90.9% (30/33) reported enhanced clinical decision-making and emergency response capabilities, 87.9% thought it helped improve medical humanistic literacy, and 87.9% expressed willingness to continue using the platform for self-directed learning.

Through systematic case design, hierarchical adaptive teaching, personalized case recommendations, and a three-stage teaching process, the AI-based intelligent case platform can significantly improve critical care medicine graduate students' clinical case analysis ability, clinical decision-making, and emergency response capabilities. Widely recognized by students, it provides an effective new teaching paradigm for graduate education in critical care medicine and has the potential to be promoted to other clinical disciplines.