Page 124 - AIH-2-1

Artificial Intelligence in Health Does improving diagnostic accuracy increase AI adoption?

These findings underscore the importance of enhancing 7. Contributions of this study
diagnostic accuracy in fostering public trust in AI-based
tools. Public acceptance of AI diagnostics is closely tied to Unlike earlier research that has largely focused on
accuracy levels, and these results suggest that AI tools must survey-based methods, this study expands the body of
meet or exceed a 95% performance threshold to achieve knowledge on AI adoption by conducting investigations
meaningful levels of AI acceptance. on AI acceptance through randomized, scenario-driven
experiments. Using this approach, we can capture a more
In addition, AI consistently outperformed AI+ across detailed perspective on how people react to AI in diverse
all levels of sensitivity, with the exception of 60% sensitivity, and controlled situations, addressing the broader challenge
where no significant difference in preference between AI of AI hesitancy and the complexity of its acceptance in
and AI+ was found. This outcome may indicate a hesitancy real-world settings. Our findings significantly enhance
towards the integration of digital consumer data -this is the current body of research by providing empirical
what AI+ means compared to AI- versus EHR data alone. evidence on the threshold of diagnostic accuracy required
However, when accuracy approaches a sensitivity level of for AI-driven technologies to achieve widespread public
95%, the public appears more willing to consider the use of acceptance. By quantifying these levels of accuracy, we offer
these digital consumer data resources, reflecting a trust deficit a framework for understanding the public’s expectations
that can be mitigated by increased diagnostic performance. of AI in health-care settings. This research not only

6. Study limitations underscores the importance of reliability and accuracy
in AI diagnostics but also highlights the nuanced factors
It is important to emphasize that, to minimize the biases influencing public trust and adoption. In addition, it sheds
of physical invasiveness while striving to level the playing light on how varying degrees of accuracy can shape public
15
field in comparison to AI testing methods, we deliberately perceptions, offering insights for developers, policymakers,
chose a salivary test for this study. As a result, our estimates and health-care professionals aiming to bridge the gap
of the public’s preference for biological tests may be in between technological advancements and public readiness
fact lower if AI testing was compared to more physically for AI integration. These insights are particularly valuable
invasive procedures such as brain imaging, cerebrospinal in addressing AI hesitancy and ensuring the ethical
fluid analysis, or blood tests. 16,17 This decision likely implementation of AI in health care.
shaped the participants’ responses, as the less invasive
nature of the salivary test may have led them to favor it 8. Conclusion
over more physically invasive testing methods. As a result, Our findings carry important implications for the
the reluctance toward AI diagnostics observed in this development and implementation of AI diagnostics in
study may be less significant when compared to scenarios health care. Public hesitation persists as a significant
involving more invasive testing procedures.
barrier, especially when AI tools are perceived as lacking
Public perceptions of AI adoption are also likely to sufficient accuracy or integrating excessive amounts of
differ significantly across geographic regions, influenced personal data. Our results emphasize the critical need
by varying cultural, economic, and social factors that shape for AI developers and health-care providers to prioritize
attitudes toward technology. Although previous studies have transparency, accuracy, and usability in AI diagnostic
shown similar AI hesitancies, this study was conducted in technologies. Moreover, educating the public about the
France and national differences could result in diverse levels potential benefits of AI diagnostics, particularly diagnostic
of trust, familiarity, and comfort with AI, thereby affecting accuracy, could further alleviate concerns and promote
how AI technologies are embraced across different nations. broader acceptance.
Consequently, this variability poses a potential limitation
to the generalizability of this study’s findings. Factors such This study highlights the nuanced preferences of the
as regional regulatory environments, access to technology, public for AI diagnostics, with higher sensitivity and
socioeconomic disparities, and historical experiences with specificity acting as key drivers of acceptance. While
digital tools could further amplify these discrepancies in AI holds considerable potential to transform health-
AI acceptance. Therefore, our findings must be considered care diagnostics, addressing the public’s concerns about
within the diverse global contexts where AI technologies accuracy and complexity will be essential to its successful
may be implemented. This underscores the importance of adoption.
future research to examine AI adoption across a broader Acknowledgments
range of geographic and cultural settings, ensuring greater
applicability and relevance. None.

Volume 2 Issue 1 (2025) 118 doi: 10.36922/aih.3561