A cognitive–motivational perspective on AI-assisted pre-clerkship preparation: a theory- informed cross-sectional study

Technical Summary: A Cognitive–Motivational Perspective on AI-Assisted Pre-Clerkship Preparation

Problem Statement
Medical education faces a critical transition gap between theoretical knowledge acquisition and experiential clinical practice, particularly in specialties like obstetrics and gynecology. Students frequently encounter structural barriers during clerkships, including restricted patient access due to privacy concerns, limited hands-on examination opportunities, and variability in clinical case exposure. These factors often lead to inadequate preparedness and a "pre-clinical cognitive scaffolding gap," where learners lack structured mechanisms to organize and activate knowledge before engaging in authentic clinical contexts. While Artificial Intelligence (AI), particularly Large Language Models (LLMs), offers potential for interactive and adaptive learning, its role in the specific preparatory phase preceding clinical exposure remains under-optimized and insufficiently understood regarding the underlying learning mechanisms and learner acceptance.

Methodology
The study employed a theory-informed cross-sectional design to examine medical students' acceptance of an AI-assisted pre-clerkship preparation model.

• Participants: 176 undergraduate medical students from Hangzhou City University (China) were surveyed between February and April 2026, prior to their obstetrics and gynecology clerkship. The effective response rate was 91.7% (176 valid responses from 192 collected).
• Instrument: A 32-item questionnaire was developed based on the Technology Acceptance Model (TAM) and literature on medical education. It assessed demographic characteristics, digital literacy, clerkship expectations, perceived usefulness (5 items), behavioral intention (willingness to participate), and implementation preferences.
• Theoretical Framework: The study was guided by TAM, positing that perceived usefulness is the primary determinant of behavioral intention. Perceived usefulness was operationalized as a composite score (mean of 5 items), and behavioral intention was measured via a single item assessing willingness to actively participate.
• Analysis: Statistical analysis included descriptive statistics, Pearson/Spearman correlation analyses, and multiple linear regression using SPSS version 26.0. Variables significant in univariate analysis ($P < 0.10$) were entered into a multivariable model to identify independent predictors.

Key Results

• Acceptance Levels: Students demonstrated moderate-to-high acceptance of AI-assisted preparation, with a mean willingness score of $3.73 \pm 0.70$. Approximately 63.0% expressed an intention to participate in such activities.
• Predictors of Engagement: In univariate analyses, perceived usefulness, academic performance, and digital literacy showed associations with behavioral intention. However, in the multivariable regression analysis, perceived usefulness emerged as the only significant independent predictor ($\beta = 0.866, P < 0.001$).
• Variance Explained: The model explained 71.8% of the variance in behavioral intention ($Adjusted\ R^2 = 0.718$).
• Non-Significant Factors: After adjustment, variables such as self-rated academic performance, theoretical course grades, digital literacy confidence, familiarity with medical databases, and career intention were not independently associated with behavioral intention. This suggests their influence may be mediated through perceived usefulness.
• Implementation Preferences: Students preferred concise, structured formats. Key preferences included:
  • Timing: 2–3 days before the clerkship (34.1%).
  • Content: Explicit learning objectives (50%) and disease lists with guiding questions (35.3%) were prioritized over virtual patient cases.
  • Format: Written summaries (33.6%) and concept maps (25.8%).
  • Duration: A strong preference for efficiency, with 69.9% favoring tasks completed within 30 minutes.

Significance and Claims
The paper claims to support a simplified Technology Acceptance Model where perceived usefulness is the primary driver of student engagement with AI-assisted pre-clerkship preparation, outweighing individual differences in digital literacy or prior academic performance.

The authors argue that these findings challenge the assumption that technical proficiency is the primary barrier to AI adoption in medical education. Instead, they propose a value-driven adoption mechanism: students engage with AI tools when they perceive clear relevance to clinical learning and performance. The study suggests that AI-assisted preparation can address the pre-clinical cognitive scaffolding gap by providing structured, cognitively supportive learning experiences that help students organize knowledge and reduce cognitive load before entering the clinic.

The authors conclude that educational implementation should prioritize meaningful learning value and clinical relevance over technological complexity. They posit that AI should be conceptualized not as a replacement for clinical teaching, but as a preparatory scaffold that enhances behavioral engagement, thereby potentially contributing to improved clinical readiness.

Limitations
The authors acknowledge several limitations: the single-institution design limits generalizability; the use of self-reported measures introduces potential response bias; behavioral intention was measured via a single item; and the study assessed anticipated engagement rather than actual usage or downstream learning outcomes. Future research is recommended to employ longitudinal and experimental designs to evaluate the impact on actual clinical performance.