Explainability in action: A metric-driven assessment of local explanations for healthcare tabular models
PLOS One · 08.07.2026 · 21 (7) · e0351473
DOI: 10.1371/journal.pone.0351473 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, online erschienen am 08.07.2026. Verfasst von M Atif Qureshi und 5 weiteren Autor:innen. Beteiligte Einrichtung: ADAPT Centre, Dublin, Ireland. Hauptthemen laut MEDLINE: Machine Learning, Delivery of Health Care. Die Arbeit nennt 59 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Research Ireland, Research Ireland Centre for Research Training in Machine Learning, Research Ireland, ADAPT Research Centre.
Abstract (Originalsprache)
Explainable AI (XAI) is increasingly used in clinical machine learning, yet quantitative evaluation of explanation quality is often reported inconsistently across methods and datasets. We present a reproducible, metric-driven framework for evaluating XAI methods on healthcare tabular data. The framework consolidates six established, family-specific metrics, fidelity, simplicity, consistency, robustness, precision, and coverage, into explicit equations; pairs them with a pre-specified focal-model protocol; and releases open-source code with a method–metric applicability map. We evaluate LIME, SHAP, Anchors, EBM, and TabNet across four public healthcare tabular datasets. Post-hoc explainers are applied to a single selected Random Forest focal predictor to control model-induced variability, whereas EBM and TabNet are assessed through their native interpretability mechanisms. Global explanation summaries are reported descriptively only. The results show that SHAP/TreeSHAP provides exact score reconstruction for the Random Forest setting, while LIME produces simpler but lower-fidelity explanations with greater instance-level variability. LIME and SHAP show the strongest rank agreement among the evaluated pairs, although agreement varies across datasets. TabNet often yields compact native explanations, but these must be interpreted alongside its dataset-specific predictive performance. EBM and TabNet show low sensitivity under the fixed Gaussian-jitter robustness protocol, while Anchors produces high-precision rules with reduced coverage at stricter thresholds. Overall, the framework enables controlled comparison under explicit method–metric and focal-model assumptions, supporting more transparent XAI selection for tabular machine learning. Although demonstrated in healthcare, the framework is transferable to other high-stakes tabular domains. Source code: https://github.com/matifq/XAI_Tab_Health .
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- ADAPT Centre, Dublin, Ireland.
- Research Ireland Centre for Research Training in Machine Learning (ML-Labs), Dublin, Ireland.
- Faculty of Business, eXplainable Analytics Group, Technological University Dublin, Dublin, Ireland.
- School of Information and Communication Studies, University College Dublin, Dublin, Ireland.
- Maynooth University, School of Business, Maynooth, Co. Kildare, Ireland.
ORCID-Profile: M Atif Qureshi, Awais Manzoor
Förderung
- Research Ireland
- Research Ireland Centre for Research Training in Machine Learning
- Research Ireland, ADAPT Research Centre
Zitieren
APA
Qureshi, M. A., Noor, A. A., Manzoor, A., Mazhar Qureshi, M. D., Younus, A., & Rashwan, W. (2026). Explainability in action: A metric-driven assessment of local explanations for healthcare tabular models. PLOS One, 21(7), e0351473. https://doi.org/10.1371/journal.pone.0351473
Vancouver
Qureshi MA, Noor AA, Manzoor A, Mazhar Qureshi MD, Younus A, Rashwan W. Explainability in action: A metric-driven assessment of local explanations for healthcare tabular models. PLOS One. 2026;21(7):e0351473. doi:10.1371/journal.pone.0351473
BibTeX
@article{Qureshi2026,
author = {Qureshi, M Atif and Noor, Abdul Aziz and Manzoor, Awais and Mazhar Qureshi, Muhammad Deedahwar and Younus, Arjumand and Rashwan, Wael},
title = {Explainability in action: A metric-driven assessment of local explanations for healthcare tabular models},
journal = {PLOS One},
year = {2026},
volume = {21},
number = {7},
pages = {e0351473},
doi = {10.1371/journal.pone.0351473}
}
EUVIMED