Adoption of AI Clinical Decision Support Systems in Primary Care: An Evaluation of Physician Acceptance
Author(s)
Jorge Brantes Ferreira1, Jorge Ferreira da Silva, PhD2, Fernanda Leao Ramos, PhD2, Cristiane Junqueira Giovannini, PhD2, Daniel Brantes Ferreira, PhD3.
1Associate Professor, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil, 2Business Administration, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil, 3Ambra University, Orlando, FL, USA.
1Associate Professor, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil, 2Business Administration, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil, 3Ambra University, Orlando, FL, USA.
OBJECTIVES: AI-based clinical decision support systems (AI-CDSS) have been increasingly deployed to assist primary care physicians in diagnosis and treatment planning. However, real-world adoption remains suboptimal. This study develops and empirically tests a structural model of physician acceptance of AI-CDSS, grounded in the Technology Acceptance Model (TAM), trust literature, and clinical autonomy frameworks.
METHODS: A cross-sectional survey was conducted with 619 Brazilian primary care physicians. The questionnaire measured constructs from TAM (Perceived Usefulness, Perceived Ease of Use, Attitude toward Use, and Behavioral Intention) plus Trust in AI and Perceived Clinical Autonomy. Structural equation modeling (SEM) was used to test the extended TAM, estimating relationships among these constructs and assessing model fit.
RESULTS: Structural equation modeling supported the hypothesized model and presented a good fit to the data. Perceived Usefulness and Perceived Ease of Use had significant positive effects on physicians’ attitude toward AI-CDSS (p < 0.01). Physicians’ attitude, in turn, significantly predicted their behavioral intention to use AI-CDSS (p < 0.001). Notably, Trust in AI had a strong positive effect on both attitude and intention (p < 0.001). Similarly, higher Perceived Clinical Autonomy was significantly associated with more positive attitudes and greater intention to use AI-CDSS (p < 0.05). Together, these constructs explained 65% of the variance in attitude and 61% of the variance in intention.
CONCLUSIONS: In this Brazilian primary care sample, physicians’ acceptance of AI-CDSS was strongly influenced by trust in AI and perceived clinical autonomy. Integrating these factors into the Technology Acceptance Model (TAM) highlights novel determinants of technology adoption. Efforts to build physician trust in AI (through transparency and training) and to ensure AI tools preserve clinical autonomy may enhance the uptake of AI-CDSS. These findings have implications for policymakers and system developers seeking to promote the effective implementation of AI-CDSS in healthcare.
METHODS: A cross-sectional survey was conducted with 619 Brazilian primary care physicians. The questionnaire measured constructs from TAM (Perceived Usefulness, Perceived Ease of Use, Attitude toward Use, and Behavioral Intention) plus Trust in AI and Perceived Clinical Autonomy. Structural equation modeling (SEM) was used to test the extended TAM, estimating relationships among these constructs and assessing model fit.
RESULTS: Structural equation modeling supported the hypothesized model and presented a good fit to the data. Perceived Usefulness and Perceived Ease of Use had significant positive effects on physicians’ attitude toward AI-CDSS (p < 0.01). Physicians’ attitude, in turn, significantly predicted their behavioral intention to use AI-CDSS (p < 0.001). Notably, Trust in AI had a strong positive effect on both attitude and intention (p < 0.001). Similarly, higher Perceived Clinical Autonomy was significantly associated with more positive attitudes and greater intention to use AI-CDSS (p < 0.05). Together, these constructs explained 65% of the variance in attitude and 61% of the variance in intention.
CONCLUSIONS: In this Brazilian primary care sample, physicians’ acceptance of AI-CDSS was strongly influenced by trust in AI and perceived clinical autonomy. Integrating these factors into the Technology Acceptance Model (TAM) highlights novel determinants of technology adoption. Efforts to build physician trust in AI (through transparency and training) and to ensure AI tools preserve clinical autonomy may enhance the uptake of AI-CDSS. These findings have implications for policymakers and system developers seeking to promote the effective implementation of AI-CDSS in healthcare.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
MT2
Topic
Medical Technologies
Topic Subcategory
Digital Health
Disease
No Additional Disease & Conditions/Specialized Treatment Areas