JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI) is increasingly being integrated into healthcare, but successful implementation depends not only on technical performance but also on stakeholder acceptance, perceived usefulness, and trust. Comparative evidence across physicians, other healthcare professionals, and the general population remains limited, particularly for patient-facing communication applications such as AI-assisted simplification of medical reports.

Objective:

This study aimed to compare AI-related knowledge, attitudes, learning readiness, and perceived need between physicians and other healthcare professionals in Austria; to examine public perceptions of AI in healthcare, particularly trust in AI-assisted diagnosis and the perceived need for simplified medical reports; and to evaluate the psychometric properties of the adapted survey instruments.

Methods:

Three parallel cross-sectional online surveys were conducted in Austria. Physicians (n=69) and other healthcare professionals (n=192) completed an identical 36-item questionnaire with six intended domains rated on 5-point Likert scales. The general population (n=106) completed a separate 37-item questionnaire rated on 4-point Likert scales. Group comparisons in the professional samples were performed using Mann-Whitney U tests with Benjamini-Hochberg correction. Internal consistency was assessed using Cronbach α. Dimensionality of the professional questionnaire was examined using confirmatory factor analysis (CFA) and exploratory factor analysis (EFA).

Results:

Physicians scored significantly higher than other healthcare professionals on 5 of 6 professional subscales, including AI Knowledge (mean 3.20 vs 2.92; adjusted P=.029; r=0.14), Attitude (mean 3.52 vs 3.11; adjusted P<.001; r=0.25), Future Expectations (mean 3.54 vs 3.22; adjusted P<.001; r=0.28), AI in Healthcare (mean 3.82 vs 3.33; adjusted P<.001; r=0.28), and Perceived Need (mean 3.39 vs 2.92; adjusted P=.001; r=0.21). Learning Readiness did not differ significantly between groups (adjusted P=.521). In both professional samples, male respondents reported higher AI Knowledge than female respondents. No significant associations with years of professional experience were observed. In the general population, the highest scores were observed for Data Sharing (mean 3.19/4.00) and Need for Simplified Medical Reports (mean 3.01/4.00), whereas Willingness to Pay was lowest (mean 2.24/4.00). CFA indicated poor fit of the intended 6-factor professional model (comparative fit index 0.780; Tucker-Lewis index 0.761; root mean square error of approximation 0.094). EFA suggested that the Future Expectations and Learning Readiness domains were multidimensional.

Conclusions:

Physicians showed higher AI-related knowledge and more positive attitudes toward AI than other healthcare professionals, whereas learning readiness was similarly high in both groups. The general population expressed a clear perceived need for simplified medical reports but lower willingness to pay for such services. These findings support the relevance of AI-assisted medical report simplification while also indicating a need for differentiated AI education across professional groups and further refinement of the adapted survey instruments. Clinical Trial: The study was reviewed and approved by the Ethics Committee of the Medical University of Graz (Reference No. 33-199 ex 20/21).