EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Human-in-the-loop artificial intelligence for stroke care: clinical applications, challenges and future directions

Gloria C. Okoye, Paul Bolaji, Nicholas Aderinto, Damilola Akinsulire, Osazuwa Ighodaro, David B. Olawade

Brain Hemorrhages · 2026-11

DOI: 10.1016/j.hest.2026.11.002 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Brain Hemorrhages, 2026-11. Verfasst von Gloria C. Okoye und 5 weiteren Autor:innen. Die Arbeit nennt 87 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Artificial intelligence (AI), including machine learning and deep learning, has shown considerable promise in stroke prediction, diagnosis, imaging interpretation, clinical decision support, and management, particularly in acute reperfusion therapy. As increasingly autonomous AI systems move toward routine use across ischaemic and haemorrhagic stroke care, concerns regarding transparency, clinical accountability, patient safety, ethical governance, and automation bias underscore the continuing importance of meaningful human oversight. Human-in-the-Loop (HITL) AI provides a collaborative framework that embeds structured clinician involvement into AI-driven decision-making. This review examines the applications, effectiveness, and clinical implications of HITL AI across the stroke care continuum, spanning primary prevention, risk prediction, emergency triage, neuroimaging interpretation, hyperacute management, rehabilitation, and secondary prevention, with attention to clinical workflows, governance, implementation challenges, and future opportunities. A narrative review methodology was adopted, drawing on MEDLINE/PubMed, EMBASE, CINAHL, and the Cochrane Library, and including studies published between 2012 and June 2026 that reported AI-assisted stroke applications with explicit human oversight. Current evidence indicates that AI systems incorporating human oversight show promising performance across these domains; however, because few studies directly compare AI alone with AI plus oversight, the specific contribution of the HITL component remains largely unestablished, and gains should be attributed to the combined human-AI system. Challenges include algorithmic bias, workflow disruption, alert fatigue, variable clinician acceptance, and limited governance. Evidence is strongest in ischaemic stroke, while haemorrhagic applications remain underdeveloped. Realising HITL AI’s potential will require co-designed workflows, robust regulation, model auditing, improved AI literacy, and prospective multicentre validation.

Zitieren

APA

Okoye, G. C., Bolaji, P., Aderinto, N., Akinsulire, D., Ighodaro, O., & Olawade, D. B. (2026). Human-in-the-loop artificial intelligence for stroke care: clinical applications, challenges and future directions. Brain Hemorrhages. https://doi.org/10.1016/j.hest.2026.11.002

Vancouver

Okoye GC, Bolaji P, Aderinto N, Akinsulire D, Ighodaro O, Olawade DB. Human-in-the-loop artificial intelligence for stroke care: clinical applications, challenges and future directions. Brain Hemorrhages. 2026. doi:10.1016/j.hest.2026.11.002

BibTeX
@article{Okoye2026,
  author = {Okoye, Gloria C. and Bolaji, Paul and Aderinto, Nicholas and Akinsulire, Damilola and Ighodaro, Osazuwa and Olawade, David B.},
  title = {Human-in-the-loop artificial intelligence for stroke care: clinical applications, challenges and future directions},
  journal = {Brain Hemorrhages},
  year = {2026},
  doi = {10.1016/j.hest.2026.11.002}
}
Kostenlos starten →