Currently submitted to: JMIR AI
Date Submitted: Jul 1, 2026
Open Peer Review Period: Jul 13, 2026 - Sep 7, 2026
(currently open for review)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Beyond Algorithm Performance: Clinical Utility and Trust Drive Physician Adoption of AI-Assisted Radiography: Cross-Sectional Survey Study
ABSTRACT
Background:
Artificial intelligence (AI) has demonstrated high diagnostic performance in radiography; however, its integration into routine clinical practice remains inconsistent.
Objective:
This study investigated the factors associated with physicians' adoption of AI-assisted radiography, focusing on perceived clinical utility, trust, prior AI experience, and recommendation behavior.
Methods:
A cross-sectional survey was conducted among physicians who used AI-assisted radiography systems in routine clinical practice. Responses were analyzed using descriptive statistics, Spearman’s correlation coefficients, and multivariable logistic regression. Composite scores were constructed for perceived usability, clinical utility, and trust. Recommendation behavior was assessed using a Net Promoter Score (NPS)-based approach.
Results:
A total of 142 physicians' responses were analyzed. Perceived clinical utility (3.36 ± 1.06) and usability (3.72 ± 1.00) were rated more favorably than trust (2.53 ± 1.06). Recommendation behavior was strongly associated with clinical utility (r = 0.71) and trust (r = 0.70), whereas prior AI experience showed only weak associations (r = 0.25). Physicians who reported AI-related errors had significantly lower trust scores (2.37 vs. 2.91; p = 0.009). In the multivariable analysis, perceived clinical utility (p = 0.001) and trust (p = 0.005) independently predicted recommendation behavior.
Conclusions:
Physician adoption of AI-assisted radiography appears to depend primarily on perceived clinical utility and trust rather than prior experience with AI systems. These findings suggest that the successful clinical implementation of medical AI will depend less on further improvements in algorithmic performance and more on trustworthy, workflow-integrated, and governance-supported human–AI collaboration.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.