Page 120 - AIH-2-1

Artificial Intelligence in Health

BRIEF REPORT
Does improving diagnostic accuracy increase

artificial intelligence adoption? A public
acceptance survey using randomized scenarios

of diagnostic methods

1,2
2
2
3
Yulin Hswen * , Ismaël Rafaï , Antoine Lacombe , Bérengère Davin-Casalena ,
Dimitri Dubois , Thierry Blayac , and Bruno Ventelou 2
4
4
1 Department of Epidemiology and Biostatistics, University of California San Francisco, San
Francisco, California, United States of America
2 Aix-Marseille Univiversity, CNRS, AMSE, Marseille, France
3 Observatoire Régional de la Santé, Provence-Alpes-Côte d’Azur, France
4 CEE-M, Univ. Montpellier, CNRS, INRAe, Institut Agro, Montpellier, France
(This article belongs to the Special Issue: Artificial intelligence for diagnosing brain diseases)

Abstract
*Corresponding author:
Yulin Hswen This study examines the acceptance of artificial intelligence (AI)-based diagnostic
(yulin.hswen@ucsf.edu) alternatives compared to traditional biological testing through a randomized
Citation: Hswen Y, Rafaï I, scenario experiment in the domain of neurodegenerative diseases (NDs). A total of
Lacombe A, et al. Does improving 3225 pairwise choices of ND risk-prediction tools were offered to participants, with
diagnostic accuracy increase 1482 choices comparing AI with the biological saliva test and 1743 comparing AI+
artificial intelligence adoption? A
public acceptance survey using with the saliva test (with AI+ using digital consumer data, in addition to electronic
randomized scenarios of diagnostic medical data). Overall, only 36.68% of responses showed preferences for AI/AI+
methods. Artif Intell Health. alternatives. Stratified by AI sensitivity levels, acceptance rates for AI/AI+ were
2025;2(1):114-120.
doi: 10.36922/aih.3561 35.04% at 60% sensitivity and 31.63% at 70% sensitivity, and increased markedly to
48.68% at 95% sensitivity (p <0.01). Similarly, acceptance rates by specificity were
Received: May 2, 2024
29.68%, 28.18%, and 44.24% at 60%, 70%, and 95% specificity, respectively (P < 0.01).
1st revised: August 2, 2024 Notably, AI consistently garnered higher acceptance rates (45.82%) than AI+
2nd revised: September 17, 2024 (28.92%) at comparable sensitivity and specificity levels, except at 60% sensitivity,
where no significant difference was observed. These results highlight the nuanced
3rd revised: September 27, 2024
preferences for AI diagnostics, with higher sensitivity and specificity significantly
Accepted: September 27, 2024 driving acceptance of AI diagnostics.
Published Online: October 18,
2024
Keywords: Artificial intelligence; AI diagnostics; Neurodegenerative diseases; Machine
Copyright: © 2024 Author(s). learning
This is an Open-Access article
distributed under the terms of the
Creative Commons Attribution
License, permitting distribution,
and reproduction in any medium, 1. Background
provided the original work is
properly cited. The integration of artificial intelligence (AI) into health care brings the promise of
Publisher’s Note: AccScience revolutionizing diagnostic and prognostic capabilities, offering more precise, data-
Publishing remains neutral with driven insights that can enhance patient outcomes. By harnessing AI’s ability to analyze
1,2
regard to jurisdictional claims in
published maps and institutional large datasets, including electronic health records (EHRs) and digital data concerning
affiliations. consumer behaviors, health-care systems can potentially improve the early detection

Volume 2 Issue 1 (2025) 114 doi: 10.36922/aih.3561