Development and application of an artificial intelligence agent-based case teaching model for health assessment: A quasi-experimental study.

BACKGROUND: Health assessment is a cornerstone of nursing education, playing a pivotal role in fostering comprehensive clinical assessment skills and clinical reasoning competence among students. However, traditional teaching faces persistent challenges: poor knowledge integration; insufficient clinical practice opportunities; and ineffective skills training. Artificial intelligence (AI) offers promise for overcoming these challenges. AIM: To develop an AI agent-based case teaching approach for the health assessment course, evaluate its effectiveness, and provide evidence-based references for AI-enabled nursing teaching practice. DESIGN: Single-center, parallel-group, quasi-randomized controlled study. SETTINGS: Southern Medical University, Guangzhou, China. PARTICIPANTS: 102 sophomore nursing students. METHODS: A dual-agent architecture was developed via workflow design by integrating large language model technology and programming code, based on a constructed knowledge base and data tables. Cluster random sampling was conducted by laboratory unit, and students were assigned to the experimental group (n = 51, AI agent-based case teaching) or the control group (n = 51, traditional platform-based teaching). Outcome measures included a personal information form, the Clinical Reasoning Scale, a teaching satisfaction questionnaire, and knowledge assessment; platform backend usage metrics were collected. Data analysis was performed using independent-samples t-tests, chi-squared tests, and linear regression models. RESULTS: The experimental group showed higher scores in clinical reasoning competence (t = 2.926, p = 0.004, Cohen's d = 0.58), knowledge (t = 2.602, p = 0.011, Cohen's d = 0.52), and teaching satisfaction (t = 3.028, p = 0.003) compared with the control group. The intervention positively influenced the test scores (β = 4.449, p = 0.0214, 95% confidence interval: [2.551, 6.347]) and clinical reasoning competence (β = 3.961, p = 0.025, 95% confidence interval: [1.956, 5.966]) of students. The experimental group had an average of 1.06 Question-and-Answer interactions and 137.27 simulated assessment interactions with the AI agents; the control group had an average of 12.18 traditional platform activities per person. CONCLUSIONS: AI agent-based case teaching was associated with enhanced knowledge, clinical reasoning competence, and teaching satisfaction among nursing students.