Skip to content
DraconiticResearch
Search Sources API
186,546 studies31 sourcesUpdated 13 min ago
PreprintCC BY-NC-NDmedRxiv

Usability and Physician Acceptance of AI in Chest Radiography: An Evaluation of the MeDeep Platform Using SUS and ATRAI-14

  • P. Shariat Rad
  • M. M. Hatami
  • M. A. Amani
  • S. Salehi1
  • M. H. Sharif
  • M. J. Gharegozlou
  • M. G. Salehi1
  1. 1Department of Radiology, School of Medicine, Kermanshah University of Medical Sciences, Kermanshah, Iran
Read full textPDFDOICite

Abstract

ObjectivesArtificial intelligence (AI) architectures demonstrate high proficiency in chest radiography interpretation, yet successful clinical integration is often hampered by usability challenges and physician skepticism. This study evaluates the ergonomic usability and professional acceptance of the MeDeep AI platform among medical practitioners to identify key drivers of clinical trust.

MethodsA multicenter, cross-sectional study was conducted across hospitals in Kermanshah, evaluating 88 medical professionals, including attending radiologists, radiology residents, internal medicine physicians, and medical externs. The System Usability Scale (SUS) was utilized to measure the platforms ergonomic functionality (n = 82), while the validated ATRAI-14 questionnaire quantified broader attitudes and trust regarding AI integration (n = 88).

ResultsThe MeDeep platform achieved a high mean SUS score of 84.93, indicating excellent usability uniformly experienced across all professional subgroups (p = 0.387). However, overarching AI acceptance and algorithmic trust varied significantly by clinical role. Radiology residents exhibited significantly higher overarching AI acceptance (p = 0.016) and algorithmic trust (p = 0.008) compared to attending radiologists. Multivariate regression demonstrated that user familiarity (Beta = 0.500, p < 0.001) and perceived system usability (Beta = 0.236, p = 0.012)--rather than years of clinical experience--were the primary independent positive predictors of AI acceptance. Furthermore, higher perceived usability directly predicted greater algorithmic trust (Beta = 0.282, p = 0.009).

ConclusionWhile technical usability is a necessary baseline for diagnostic platforms, professional trust dictates ultimate clinical adoption. Overcoming the generational divide in AI acceptance requires developers to prioritize intuitive interface design and position AI as a workflow-augmenting tool that addresses administrative burdens.

Key PointsO_LIThe MeDeep AI platform demonstrated excellent ergonomic usability uniformly across all professional clinical ranks. C_LIO_LISignificant generational divides exist, with residents exhibiting higher algorithmic trust than attending radiologists. C_LIO_LIUser familiarity and perceived usability predict overarching AI acceptance better than years of clinical experience. C_LI

Related

  1. Usability and Acceptance of a Social Robot in People With Multiple Sclerosis: Comparative Cross-Sectional Study.
    Jessica Podda, Alice Bollini, Alessia Susini +7
    JMIR rehabilitation and assistive technologiesPeer-reviewedCC BYFull textGeneral medicineEurope PMCDOAJ
  2. Feasibility, acceptability and usability of InControl web-platform intervention for older adults with chronic musculoskeletal pain: A mixed-methods pilot study protocol
    David Oliveira, Nelson P. Rocha, Anabela G. Silva
    PLOS OnePeer-reviewedCC BY 4.0Full textClinical trials & evidence synthesisPLOS
  3. A theory-driven mobile application based on iterative collaborative design principles to support depression self-management in older adults: development and usability evaluation of “Mind Garden”
    Ziping Zhu, Zhi Chen, Zhengmin Wang +4
    BMC Public HealthPeer-reviewedCC BY-NC-ND 4.0Epidemiology & public healthCrossref
  4. Just One Move: co-design and usability testing of a digital physical activity behavior change platform for people with rheumatoid arthritis
    Manuel Ester, Daniel Gillespie, Kiran Dhiman +11
    Frontiers in Public HealthPeer-reviewedCC BY 4.0Epidemiology & public healthCrossref
  5. Durability of a physician-modified endograft constructed using the Terumo Aortic TREO device: A postmortem evaluation
    Journal of vascular surgery cases and innovative techniquesPeer-reviewedCC BY-NC-NDFull textSurgery & anaesthesiaEurope PMC

Cite

BibTeXRIS
BibTeX
@misc{rad2026usability,
  title = {{Usability and Physician Acceptance of AI in Chest Radiography: An Evaluation of the MeDeep Platform Using SUS and ATRAI-14}},
  author = {P. Shariat Rad and M. M. Hatami and M. A. Amani and S. Salehi and M. H. Sharif and M. J. Gharegozlou and M. G. Salehi},
  howpublished = {medRxiv},
  year = {2026},
  month = oct,
  doi = {10.64898/2026.10.23.26365338},
  url = {https://www.medrxiv.org/content/10.64898/2026.10.23.26365338v1},
  note = {CC BY-NC-ND},
}
RIS
TY  - UNPB
TI  - Usability and Physician Acceptance of AI in Chest Radiography: An Evaluation of the MeDeep Platform Using SUS and ATRAI-14
AU  - P. Shariat Rad
AU  - M. M. Hatami
AU  - M. A. Amani
AU  - S. Salehi
AU  - M. H. Sharif
AU  - M. J. Gharegozlou
AU  - M. G. Salehi
PY  - 2026
DA  - 2026/10/23
PB  - medRxiv
DO  - 10.64898/2026.10.23.26365338
UR  - https://www.medrxiv.org/content/10.64898/2026.10.23.26365338v1
AB  - ObjectivesArtificial intelligence (AI) architectures demonstrate high proficiency in chest radiography interpretation, yet successful clinical integration is often hampered by usability challenges and physician skepticism. This study evaluates the ergonomic usability and professional acceptance of the MeDeep AI platform among medical practitioners to identify key drivers of clinical trust.  MethodsA multicenter, cross-sectional study was conducted across hospitals in Kermanshah, evaluating 88 medical professionals, including attending radiologists, radiology residents, internal medicine physicians, and medical externs. The System Usability Scale (SUS) was utilized to measure the platforms ergonomic functionality (n = 82), while the validated ATRAI-14 questionnaire quantified broader attitudes and trust regarding AI integration (n = 88).  ResultsThe MeDeep platform achieved a high mean SUS score of 84.93, indicating excellent usability uniformly experienced across all professional subgroups (p = 0.387). However, overarching AI acceptance and algorithmic trust varied significantly by clinical role. Radiology residents exhibited significantly higher overarching AI acceptance (p = 0.016) and algorithmic trust (p = 0.008) compared to attending radiologists. Multivariate regression demonstrated that user familiarity (Beta = 0.500, p < 0.001) and perceived system usability (Beta = 0.236, p = 0.012)--rather than years of clinical experience--were the primary independent positive predictors of AI acceptance. Furthermore, higher perceived usability directly predicted greater algorithmic trust (Beta = 0.282, p = 0.009).  ConclusionWhile technical usability is a necessary baseline for diagnostic platforms, professional trust dictates ultimate clinical adoption. Overcoming the generational divide in AI acceptance requires developers to prioritize intuitive interface design and position AI as a workflow-augmenting tool that addresses administrative burdens.  Key PointsO_LIThe MeDeep AI platform demonstrated excellent ergonomic usability uniformly across all professional clinical ranks. C_LIO_LISignificant generational divides exist, with residents exhibiting higher algorithmic trust than attending radiologists. C_LIO_LIUser familiarity and perceived usability predict overarching AI acceptance better than years of clinical experience. C_LI
ER  -