Evidence-based XAI : An empirical approach to design more effective and explainable decision support systems
Copyright © 2024 The Author(s). Published by Elsevier Ltd.. All rights reserved..
This paper proposes a user study aimed at evaluating the impact of Class Activation Maps (CAMs) as an eXplainable AI (XAI) method in a radiological diagnostic task, the detection of thoracolumbar (TL) fractures from vertebral X-rays. In particular, we focus on two oft-neglected features of CAMs, that is granularity and coloring, in terms of what features, lower-level vs higher-level, should the maps highlight and adopting which coloring scheme, to bring better impact to the decision-making process, both in terms of diagnostic accuracy (that is effectiveness) and of user-centered dimensions, such as perceived confidence and utility (that is satisfaction), depending on case complexity, AI accuracy, and user expertise. Our findings show that lower-level features CAMs, which highlight more focused anatomical landmarks, are associated with higher diagnostic accuracy than higher-level features CAMs, particularly among experienced physicians. Moreover, despite the intuitive appeal of semantic CAMs, traditionally colored CAMs consistently yielded higher diagnostic accuracy across all groups. Our results challenge some prevalent assumptions in the XAI field and emphasize the importance of adopting an evidence-based and human-centered approach to design and evaluate AI- and XAI-assisted diagnostic tools. To this aim, the paper also proposes a hierarchy of evidence framework to help designers and practitioners choose the XAI solutions that optimize performance and satisfaction on the basis of the strongest evidence available or to focus on the gaps in the literature that need to be filled to move from opinionated and eminence-based research to one more based on empirical evidence and end-user work and preferences.
Medienart: |
E-Artikel |
---|
Erscheinungsjahr: |
2024 |
---|---|
Erschienen: |
2024 |
Enthalten in: |
Zur Gesamtaufnahme - volume:170 |
---|---|
Enthalten in: |
Computers in biology and medicine - 170(2024) vom: 01. Feb., Seite 108042 |
Sprache: |
Englisch |
---|
Beteiligte Personen: |
Famiglini, Lorenzo [VerfasserIn] |
---|
Links: |
---|
Themen: |
Activation maps |
---|
Anmerkungen: |
Date Completed 28.02.2024 Date Revised 28.02.2024 published: Print-Electronic Citation Status MEDLINE |
---|
doi: |
10.1016/j.compbiomed.2024.108042 |
---|
funding: |
|
---|---|
Förderinstitution / Projekttitel: |
|
PPN (Katalog-ID): |
NLM367983451 |
---|
LEADER | 01000caa a22002652 4500 | ||
---|---|---|---|
001 | NLM367983451 | ||
003 | DE-627 | ||
005 | 20240229170841.0 | ||
007 | cr uuu---uuuuu | ||
008 | 240204s2024 xx |||||o 00| ||eng c | ||
024 | 7 | |a 10.1016/j.compbiomed.2024.108042 |2 doi | |
028 | 5 | 2 | |a pubmed24n1310.xml |
035 | |a (DE-627)NLM367983451 | ||
035 | |a (NLM)38308866 | ||
035 | |a (PII)S0010-4825(24)00126-4 | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
100 | 1 | |a Famiglini, Lorenzo |e verfasserin |4 aut | |
245 | 1 | 0 | |a Evidence-based XAI |b An empirical approach to design more effective and explainable decision support systems |
264 | 1 | |c 2024 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a ƒaComputermedien |b c |2 rdamedia | ||
338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
500 | |a Date Completed 28.02.2024 | ||
500 | |a Date Revised 28.02.2024 | ||
500 | |a published: Print-Electronic | ||
500 | |a Citation Status MEDLINE | ||
520 | |a Copyright © 2024 The Author(s). Published by Elsevier Ltd.. All rights reserved. | ||
520 | |a This paper proposes a user study aimed at evaluating the impact of Class Activation Maps (CAMs) as an eXplainable AI (XAI) method in a radiological diagnostic task, the detection of thoracolumbar (TL) fractures from vertebral X-rays. In particular, we focus on two oft-neglected features of CAMs, that is granularity and coloring, in terms of what features, lower-level vs higher-level, should the maps highlight and adopting which coloring scheme, to bring better impact to the decision-making process, both in terms of diagnostic accuracy (that is effectiveness) and of user-centered dimensions, such as perceived confidence and utility (that is satisfaction), depending on case complexity, AI accuracy, and user expertise. Our findings show that lower-level features CAMs, which highlight more focused anatomical landmarks, are associated with higher diagnostic accuracy than higher-level features CAMs, particularly among experienced physicians. Moreover, despite the intuitive appeal of semantic CAMs, traditionally colored CAMs consistently yielded higher diagnostic accuracy across all groups. Our results challenge some prevalent assumptions in the XAI field and emphasize the importance of adopting an evidence-based and human-centered approach to design and evaluate AI- and XAI-assisted diagnostic tools. To this aim, the paper also proposes a hierarchy of evidence framework to help designers and practitioners choose the XAI solutions that optimize performance and satisfaction on the basis of the strongest evidence available or to focus on the gaps in the literature that need to be filled to move from opinionated and eminence-based research to one more based on empirical evidence and end-user work and preferences | ||
650 | 4 | |a Journal Article | |
650 | 4 | |a Activation maps | |
650 | 4 | |a Evidence-based design | |
650 | 4 | |a Explainable AI | |
650 | 4 | |a Human-centered AI | |
650 | 4 | |a Human–AI collaboration | |
650 | 4 | |a Medical imaging | |
700 | 1 | |a Campagner, Andrea |e verfasserin |4 aut | |
700 | 1 | |a Barandas, Marilia |e verfasserin |4 aut | |
700 | 1 | |a La Maida, Giovanni Andrea |e verfasserin |4 aut | |
700 | 1 | |a Gallazzi, Enrico |e verfasserin |4 aut | |
700 | 1 | |a Cabitza, Federico |e verfasserin |4 aut | |
773 | 0 | 8 | |i Enthalten in |t Computers in biology and medicine |d 1970 |g 170(2024) vom: 01. Feb., Seite 108042 |w (DE-627)NLM000382272 |x 1879-0534 |7 nnns |
773 | 1 | 8 | |g volume:170 |g year:2024 |g day:01 |g month:02 |g pages:108042 |
856 | 4 | 0 | |u http://dx.doi.org/10.1016/j.compbiomed.2024.108042 |3 Volltext |
912 | |a GBV_USEFLAG_A | ||
912 | |a GBV_NLM | ||
951 | |a AR | ||
952 | |d 170 |j 2024 |b 01 |c 02 |h 108042 |