Abstract
Artificial Intelligence (AI) tools have shown promising results in reducing overcrowding in the emergency department (ED). Integrating a retrospectively validated AI model into clinical practice remains challenging. In this study, an AI-based admission prediction model is prospectively evaluated during its integration into the ED workflow. Clinical relevance is defined as a 10-minute reduction in the admission decision workflow. A prospective study was conducted using an iterative implementation design with sequential study periods. After each period, an evaluation was carried out with key stakeholders to identify the adjustments needed to ensure successful adoption. At the end of the study periods, the CFIR framework was used to determine implementation components that may inform further adoption. The three study periods did not shown clinically significant reduction. A 5-minute decrease was observed in the final period. Despite improvements in visibility and training, user engagement remained limited. A CFIR-based analysis helped identify barriers. These findings highlight that technical performance alone is not sufficient. Effective implementation requires clinical alignment.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/HAIC26_006.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=atBtns3ZQI
BibTex
@InProceedings{RosWie_Why_MICCAISAT2026,
author = { Roskamp, Wiesje AND van der Haas, Yvette J. AND Plas, Rogier L. C. AND Overbeek, Bart AND Vreeburg, Marleen AND de Jong, Annemarie AND de Carvalho, Renata M. AND van Dongen, Boudewijn F. AND van Dijk, Thomas},
title = { { Why AI Fails to Influence Admission Decisions: Evidence from a Case Study } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17279},
month = {pending},
page = {pending}
}