Accepted for/Published in: Journal of Medical Internet Research
Date Submitted: May 20, 2022
Date Accepted: Nov 30, 2022
(closed for review but you can still tweet)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Multi-stakeholder perspectives of clinical artificial intelligence implementation: a systematic review of qualitative evidence
ABSTRACT
Background:
The rhetoric surrounding clinical artificial intelligence (AI) exaggerates its impact on real-world care. Limited understanding of factors influencing its implementation perpetuates this.
Objective:
This qualitative systematic review aims to identify key stakeholders, consolidate their perspectives on clinical AI implementation and characterise evidence gaps which future qualitative research should target.
Methods:
Ovid-MEDLINE, EBSCO-CINAHL, ACM Digital Library, Science Citation Index-Web of Science and Scopus were searched for primary qualitative studies of any individuals’ perspectives on any application of clinical AI worldwide (January 2014-April 2021). Language was not an exclusion criterion. Two independent reviewers performed title, abstract and full-text screening and quality appraisal, with a third arbiter of disagreements. A single reviewer extracted free-text data relevant to clinical AI implementation, noting the stakeholder contributing each excerpt. Best-fit framework synthesis used the Non-adoption, Abandonment, Sustainability, Spread and Scale-up (NASSS) framework. To validate the data and improve accessibility, co-authors representing each emergent stakeholder group co-developed summaries of factors most relevant to their stakeholder group.
Results:
The initial search yielded 4437 deduplicated articles, with 111 eligible for inclusion (median quality score 8/10). Five distinct stakeholder groups emerged from the data; healthcare professionals (HCPs), patients/carers/public, developers, healthcare managers/leaders and regulators/policy makers contributing 70.0%,11.4%,7.7%,7.5% and 3.4% of eligible excerpts respectively. All stakeholder groups independently identified a breadth of implementation factors, with each producing data that mapped to between 17 and 24 of the 27 adapted NASSS subdomains.
Conclusions:
Clinical AI implementation is influenced by many interdependent factors, which vary between different groups of stakeholders. This implies that although HCP perspectives have dominated the literature, optimising clinical AI implementation will require consideration of all stakeholder groups’ perspectives. Future research should not only widen the representation of tools and contexts in qualitative research, but specifically investigate the perspectives of stakeholders besides HCPs. Clinical Trial: An a priori protocol is registered on PROSPERO (ID: CRD42021256005) and published in JMIR Research Protocols (DOI: 10.2196/33145).
Citation
