volume 138 pages 102506

Rams, hounds and white boxes: Investigating human–AI collaboration protocols in medical diagnosis

Federico Cabitza 1, 2
Andrea Campagner 2
Luca Ronzio 3
Matteo Cameli 4
Giulia E. Mandoli 4
Duarte Folgado 7
Marília Barandas 7
Hugo Gamboa 7, 8
Publication typeJournal Article
Publication date2023-04-01
scimago Q1
wos Q1
SJR1.396
CiteScore12.2
Impact factor6.2
ISSN09333657, 18732860
Medicine (miscellaneous)
Artificial Intelligence
Abstract
In this paper, we study human-AI collaboration protocols, a design-oriented construct aimed at establishing and evaluating how humans and AI can collaborate in cognitive tasks. We applied this construct in two user studies involving 12 specialist radiologists (the knee MRI study) and 44 ECG readers of varying expertise (the ECG study), who evaluated 240 and 20 cases, respectively, in different collaboration configurations. We confirm the utility of AI support but find that XAI can be associated with a “white-box paradox”, producing a null or detrimental effect. We also find that the order of presentation matters: AI-first protocols are associated with higher diagnostic accuracy than human-first protocols, and with higher accuracy than both humans and AI alone. Our findings identify the best conditions for AI to augment human diagnostic skills, rather than trigger dysfunctional responses and cognitive biases that can undermine decision effectiveness.
Found 
Found 

Top-30

Journals

1
2
3
4
5
6
7
Communications in Computer and Information Science
7 publications, 9.59%
Lecture Notes in Computer Science
5 publications, 6.85%
npj Digital Medicine
5 publications, 6.85%
International Journal of Human-Computer Interaction
4 publications, 5.48%
Artificial Intelligence in Medicine
3 publications, 4.11%
International Journal of Human Computer Studies
3 publications, 4.11%
Computer Supported Cooperative Work
2 publications, 2.74%
IEEE Access
2 publications, 2.74%
International Journal of Medical Informatics
2 publications, 2.74%
Machine Learning and Knowledge Extraction
1 publication, 1.37%
Clinical Chemistry and Laboratory Medicine
1 publication, 1.37%
Digestive and Liver Disease
1 publication, 1.37%
Mayo Clinic Proceedings Digital Health
1 publication, 1.37%
Journal of Medical Internet Research
1 publication, 1.37%
Computers in Biology and Medicine
1 publication, 1.37%
Information Fusion
1 publication, 1.37%
World Journal of Gastrointestinal Endoscopy
1 publication, 1.37%
Behaviour and Information Technology
1 publication, 1.37%
Computer Law and Security Review
1 publication, 1.37%
Handbook of Research on Teaching Methods in Language Translation and Interpretation
1 publication, 1.37%
Investigative Ophthalmology and Visual Science
1 publication, 1.37%
BMC Musculoskeletal Disorders
1 publication, 1.37%
Patterns
1 publication, 1.37%
BioMedInformatics
1 publication, 1.37%
Health Informatics Journal
1 publication, 1.37%
European Journal of Nuclear Medicine and Molecular Imaging
1 publication, 1.37%
Insights into Imaging
1 publication, 1.37%
The Future of Responsible Management Education
1 publication, 1.37%
Gastroenterology
1 publication, 1.37%
1
2
3
4
5
6
7

Publishers

5
10
15
20
25
30
Springer Nature
30 publications, 41.1%
Elsevier
18 publications, 24.66%
Association for Computing Machinery (ACM)
6 publications, 8.22%
Taylor & Francis
5 publications, 6.85%
Institute of Electrical and Electronics Engineers (IEEE)
3 publications, 4.11%
MDPI
2 publications, 2.74%
Walter de Gruyter
1 publication, 1.37%
JMIR Publications
1 publication, 1.37%
Baishideng Publishing Group
1 publication, 1.37%
IGI Global
1 publication, 1.37%
Association for Research in Vision and Ophthalmology (ARVO)
1 publication, 1.37%
SAGE
1 publication, 1.37%
Proceedings of the National Academy of Sciences (PNAS)
1 publication, 1.37%
Cold Spring Harbor Laboratory
1 publication, 1.37%
Mark Allen Group
1 publication, 1.37%
5
10
15
20
25
30
  • We do not take into account publications without a DOI.
  • Statistics recalculated weekly.

Are you a researcher?

Create a profile to get free access to personal recommendations for colleagues and new articles.
Metrics
74
Share
Cite this
GOST |
Cite this
GOST Copy
Cabitza F. et al. Rams, hounds and white boxes: Investigating human–AI collaboration protocols in medical diagnosis // Artificial Intelligence in Medicine. 2023. Vol. 138. p. 102506.
GOST all authors (up to 50) Copy
Cabitza F., Campagner A., Ronzio L., Cameli M., Mandoli G. E., Pastore M., Sconfienza L. M., Folgado D., Barandas M., Gamboa H. Rams, hounds and white boxes: Investigating human–AI collaboration protocols in medical diagnosis // Artificial Intelligence in Medicine. 2023. Vol. 138. p. 102506.
RIS |
Cite this
RIS Copy
TY - JOUR
DO - 10.1016/j.artmed.2023.102506
UR - https://doi.org/10.1016/j.artmed.2023.102506
TI - Rams, hounds and white boxes: Investigating human–AI collaboration protocols in medical diagnosis
T2 - Artificial Intelligence in Medicine
AU - Cabitza, Federico
AU - Campagner, Andrea
AU - Ronzio, Luca
AU - Cameli, Matteo
AU - Mandoli, Giulia E.
AU - Pastore, Maria
AU - Sconfienza, Luca Maria
AU - Folgado, Duarte
AU - Barandas, Marília
AU - Gamboa, Hugo
PY - 2023
DA - 2023/04/01
PB - Elsevier
SP - 102506
VL - 138
PMID - 36990586
SN - 0933-3657
SN - 1873-2860
ER -
BibTex
Cite this
BibTex (up to 50 authors) Copy
@article{2023_Cabitza,
author = {Federico Cabitza and Andrea Campagner and Luca Ronzio and Matteo Cameli and Giulia E. Mandoli and Maria Pastore and Luca Maria Sconfienza and Duarte Folgado and Marília Barandas and Hugo Gamboa},
title = {Rams, hounds and white boxes: Investigating human–AI collaboration protocols in medical diagnosis},
journal = {Artificial Intelligence in Medicine},
year = {2023},
volume = {138},
publisher = {Elsevier},
month = {apr},
url = {https://doi.org/10.1016/j.artmed.2023.102506},
pages = {102506},
doi = {10.1016/j.artmed.2023.102506}
}