The Ethics of Autonomous Clinical AI: Shifting From Model Accuracy to Model Impact
Journal of Medical Internet Research · 2026-10-23 · 28 · e112246-e112246
DOI: 10.2196/112246 ↗At a glance
Journal article in Journal of Medical Internet Research, published online on 2026-10-23. Written by Clair Sullivan. Institution: Queensland Digital Health Centre (QDHeC), University of Queensland, Level 5, UQ Health Sciences Building, Brisbane, Queensland, 4006, Australia, 61 0…. Major MEDLINE topics: Artificial Intelligence. 7 references. The full text is freely available in PubMed Central.
Topics: Ethik in der Pflege
Abstract (original language)
The ethics of autonomous clinical AI are controversial. Traditional ethics assessments have focused on AI model performance rather than health system outcomes. Nichol and colleagues translate established research ethics principles into 9 practical recommendations for trials of autonomous clinical AI. Their findings support an evolution in clinical trial design. The accuracy and performance of the model is important, but the ethical unit of evaluation should be the complete care pathway, with prespecified obligations for subgroup performance, follow-up care, system failures, and model change.
Subjects
MeSH terms (MEDLINE)
Author keywords
Authors and institutions
- Queensland Digital Health Centre (QDHeC), University of Queensland, Level 5, UQ Health Sciences Building, Brisbane, Queensland, 4006, Australia, 61 0400669296.
Cite
APA
Sullivan, C. (2026). The Ethics of Autonomous Clinical AI: Shifting From Model Accuracy to Model Impact. Journal of Medical Internet Research, 28, e112246-e112246. https://doi.org/10.2196/112246
Vancouver
Sullivan C. The Ethics of Autonomous Clinical AI: Shifting From Model Accuracy to Model Impact. Journal of Medical Internet Research. 2026;28:e112246-e112246. doi:10.2196/112246
BibTeX
@article{Sullivan2026,
author = {Sullivan, Clair},
title = {The Ethics of Autonomous Clinical AI: Shifting From Model Accuracy to Model Impact},
journal = {Journal of Medical Internet Research},
year = {2026},
volume = {28},
pages = {e112246--e112246},
doi = {10.2196/112246}
}
EUVIMED