CUE-X: A Framework for the Automatic Evaluation of Clinical Usefulness of Explanations for the Multimorbidity Problem

  • Martin Michalowski
  • , Szymon Wilk
  • , Jenny M. Bauer
  • , Marc Carrier
  • , Herna Viktor
  • , Wojtek Michalowski

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper introduces CUE-X, a novel evaluation framework designed to automatically assess the clinical usefulness of explanations generated by decision support systems for managing multimorbidity. CUE-X uses three evaluation dimensions: patient case relevance, evidence support, and comprehension. A prospective evaluation study involving Canadian and American physicians demonstrated the framework’s feasibility and showed alignment between CUE-X evaluations and physician assessments. These findings contribute to research on integrating human-centered evaluation metrics into the XAI field, paving the way for improving the uptake of decision support systems in clinical practice.

Original languageEnglish (US)
Title of host publicationArtificial Intelligence in Medicine - 23rd International Conference, AIME 2025, Proceedings
EditorsRiccardo Bellazzi, Lucia Sacchi, José Manuel Juarez Herrero, Blaž Zupan
PublisherSpringer Science and Business Media Deutschland GmbH
Pages293-303
Number of pages11
ISBN (Print)9783031958373
DOIs
StatePublished - 2025
Event23rd International Conference on Artificial Intelligence in Medicine, AIME 2025 - Pavia, Italy
Duration: Jun 23 2025Jun 26 2025

Publication series

NameLecture Notes in Computer Science
Volume15734 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference23rd International Conference on Artificial Intelligence in Medicine, AIME 2025
Country/TerritoryItaly
CityPavia
Period6/23/256/26/25

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Fingerprint

Dive into the research topics of 'CUE-X: A Framework for the Automatic Evaluation of Clinical Usefulness of Explanations for the Multimorbidity Problem'. Together they form a unique fingerprint.

Cite this