Differential analysis of artificial intelligence teaching tools acceptance among medical teachers based on theory of planned behavior

A cross-sectional study with latent profile and multi-class logistic regression analysis

Bibliographic Data

ID22425098
AuthorsRui Liu (0000-0002-1489-731X, Panzhihua University), Min Li (0000-0002-1163-8961, Dali University), Hui Ji (0000-0002-1470-1794, Panzhihua University, corresponding author)
Year2026
Volume26
Issue1
Publication date2026-04-20
Open AccessYes
TypeARTICLE
VenueBMC Medical Education (JOURNAL)
Journal identifiersISSN: 1472-6920 • E-ISSN: 1472-6920
PublisherSpringer Science and Business Media LLC (PUBLISHER)
DOI10.1186/s12909-026-09263-y
PMID42010617
OpenAlexW7154973608
LanguageEN
References cited37

BACKGROUND: The integration of AI teaching tools into medical education presents transformative opportunities, yet the acceptance by medical teachers in resource-constrained regions of China remains pivotal for its adoption. OBJECTIVE: To explore the latent profiles and influencing factors of AI teaching tools acceptance among medical teachers in these regions based on the TPB, and to provide evidence for formulating differentiated AI education promotion strategies in medical fields. METHODS: A cross-sectional study was conducted among 220 medical teachers from 10 medical universities and affiliated hospitals in southwestern Sichuan and northwestern China between May and June 2025 using mixed non-probability sampling. Data were collected through a general information questionnaire, the TAAIS, and the GAAIS. LPA and multi-class logistic regression were employed to identify potential profiles and analyze influencing factors. RESULTS: The optimal three-profile model identified distinct medical teacher acceptance profiles: Ambivalent-Hesitant (34.5%), Conservative-Skeptical (45.9%), and Proactive-Accepting (19.5%). Within the TPB framework, behavioral attitude showed the strongest association with profile membership; each one-point increase was associated with lower odds of belonging to the Ambivalent-Hesitant (OR = 0.535, P < 0.001) or Conservative-Skeptical (OR = 0.588, P < 0.001) than to the Proactive-Accepting. Subjective norms were more strongly associated with the Proactive-Accepting than with the Ambivalent-Hesitant. Perceived behavioral control was also associated with profile differentiation, reflected in significant self-efficacy differences across profiles (H = 116.451, P < 0.001) and in its association with behavioral intention. Teaching subject was additionally associated with acceptance heterogeneity (P < 0.050). CONCLUSION: Acceptance of AI teaching tools among medical teachers in resource-constrained regions clustered into three latent profiles. Profile membership was associated with behavioral attitudes, subjective norms, and perceived behavioral control. Effective integration may therefore benefit from profile-specific strategies tailored to local resources, training capacity, and institutional culture. Longitudinal studies are warranted because our cross-sectional data cannot determine whether teachers transition between these profiles over time as experience with AI tools and institutional support accumulate. Such designs would help clarify adoption trajectories and the role of organizational conditions.

Logistic regression · Odds · Structural equation modeling · Survey data collection · Theory of planned behavior · AI in Service Interactions · Artificial Intelligence in Healthcare and Education · Explainable Artificial Intelligence (XAI

Tools

Open DOIOpen Access