Explainability in action: A metric-driven assessment of local explanations for healthcare tabular models

M Atif Qureshi, Abdul Aziz Noor, Awais Manzoor, Muhammad Deedahwar Mazhar Qureshi, Arjumand Younus, Wael Rashwan · 2026

Read Explainability in action: A metric-driven assessment of local explanations for healthcare tabular models by M Atif Qureshi, Abdul Aziz Noor, Awais Manzoor, Muhammad Deedahwar Mazhar Qureshi, Arjumand Younus, Wael Rashwan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0351473.txt.

Description

M. Atif Qureshi and five colleagues present this 2026 PLOS One framework for judging whether explanations of medical machine learning models are actually any good. Clinicians increasingly receive reasons for a model's prediction, but studies measure explanation quality in inconsistent ways. The authors define six quality measures with explicit formulas and test five explanation methods on four public healthcare datasets.

One method reconstructed the model's scores exactly but produced complex explanations, while another gave simpler explanations that matched the model less faithfully and varied more between patients. Rule-based explanations were precise but covered few cases at strict settings. The authors release their code and a guide matching methods to measures, aiming to make explanation claims in high-stakes fields easier to check.

Catalog record

CREATORM Atif Qureshi, Abdul Aziz Noor, Awais Manzoor, Muhammad Deedahwar Mazhar Qureshi, Arjumand Younus, Wael Rashwan
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE96026
LICENSECC_BY
RIGHTSCC BY
SOURCEPLOS

Source: https://doi.org/10.1371/journal.pone.0351473

Browse the full library · About the Tower of Babel