Non-task Expert Physicians Benefit from Correct Explainable AI Advice when Reviewing X-rays

Overview

Journal Sci Rep

Specialty Science

Date 2023 Jan 25

PMID 36697450

Authors

Susanne Gaube

Harini Suresh

Martina Raue

Eva Lermer

Timo K Koch

Matthias F C Hudecek

Alun D Ackery

Samir C Grover

Joseph F Coughlin

Dieter Frey

Felipe C Kitamura

Marzyeh Ghassemi

Errol Colak

Affiliations

Soon will be listed here.

Abstract

Artificial intelligence (AI)-generated clinical advice is becoming more prevalent in healthcare. However, the impact of AI-generated advice on physicians' decision-making is underexplored. In this study, physicians received X-rays with correct diagnostic advice and were asked to make a diagnosis, rate the advice's quality, and judge their own confidence. We manipulated whether the advice came with or without a visual annotation on the X-rays, and whether it was labeled as coming from an AI or a human radiologist. Overall, receiving annotated advice from an AI resulted in the highest diagnostic accuracy. Physicians rated the quality of AI advice higher than human advice. We did not find a strong effect of either manipulation on participants' confidence. The magnitude of the effects varied between task experts and non-task experts, with the latter benefiting considerably from correct explainable AI advice. These findings raise important considerations for the deployment of diagnostic advice in healthcare.

Citing Articles

Unveiling GPT-4V's hidden challenges behind high accuracy on USMLE questions: Observational Study.

Yang Z, Yao Z, Tasmin M, Vashisht P, Jang W, Ouyang F J Med Internet Res. 2025; 27:e65146.

PMID: 39919278 PMC: 11845889. DOI: 10.2196/65146.

Explainable AI improves task performance in human-AI collaboration.

Senoner J, Schallmoser S, Kratzwald B, Feuerriegel S, Netland T Sci Rep. 2024; 14(1):31150.

PMID: 39730794 PMC: 11681242. DOI: 10.1038/s41598-024-82501-9.

Care to Explain? AI Explanation Types Differentially Impact Chest Radiograph Diagnostic Performance and Physician Trust in AI.

Prinster D, Mahmood A, Saria S, Jeudy J, Lin C, Yi P Radiology. 2024; 313(2):e233261.

PMID: 39560483 PMC: 11605106. DOI: 10.1148/radiol.233261.

How Explainable Artificial Intelligence Can Increase or Decrease Clinicians' Trust in AI Applications in Health Care: Systematic Review.

Rosenbacke R, Melhus A, Mckee M, Stuckler D JMIR AI. 2024; 3:e53207.

PMID: 39476365 PMC: 11561425. DOI: 10.2196/53207.

Advice from artificial intelligence: a review and practical implications.

Baines J, Dalal R, Ponce L, Tsai H Front Psychol. 2024; 15:1390182.

PMID: 39439754 PMC: 11493662. DOI: 10.3389/fpsyg.2024.1390182.

References

Lee H, Yune S, Mansouri M, Kim M, Tajmir S, Guerrier C . An explainable deep-learning algorithm for the detection of acute intracranial haemorrhage from small datasets. Nat Biomed Eng. 2019; 3(3):173-182. DOI: 10.1038/s41551-018-0324-9. View

van Leeuwen K, Schalekamp S, Rutten M, van Ginneken B, de Rooij M . Artificial intelligence in radiology: 100 commercially available products and their scientific evidence. Eur Radiol. 2021; 31(6):3797-3804. PMC: 8128724. DOI: 10.1007/s00330-021-07892-z. View

Killock D . AI outperforms radiologists in mammographic screening. Nat Rev Clin Oncol. 2020; 17(3):134. DOI: 10.1038/s41571-020-0329-7. View

Ozuru Y, Briner S, Kurby C, McNamara D . Comparing comprehension measured by multiple-choice and open-ended questions. Can J Exp Psychol. 2013; 67(3):215-27. DOI: 10.1037/a0032918. View

Li D, Pehrson L, Lauridsen C, Tottrup L, Fraccaro M, Elliott D . The Added Effect of Artificial Intelligence on Physicians' Performance in Detecting Thoracic Pathologies on CT and Chest X-ray: A Systematic Review. Diagnostics (Basel). 2021; 11(12). PMC: 8700414. DOI: 10.3390/diagnostics11122206. View

Raoof S, Feigin D, Sung A, Raoof S, Irugulpati L, Rosenow 3rd E . Interpretation of plain chest roentgenogram. Chest. 2012; 141(2):545-558. DOI: 10.1378/chest.10-1302. View

Rajpurkar P, Irvin J, Ball R, Zhu K, Yang B, Mehta H . Deep learning for chest radiograph diagnosis: A retrospective comparison of the CheXNeXt algorithm to practicing radiologists. PLoS Med. 2018; 15(11):e1002686. PMC: 6245676. DOI: 10.1371/journal.pmed.1002686. View

Tariq A, Purkayastha S, Padmanaban G, Krupinski E, Trivedi H, Banerjee I . Current Clinical Applications of Artificial Intelligence in Radiology and Their Best Supporting Evidence. J Am Coll Radiol. 2020; 17(11):1371-1381. DOI: 10.1016/j.jacr.2020.08.018. View

Geijer H, Geijer M . Added value of double reading in diagnostic radiology,a systematic review. Insights Imaging. 2018; 9(3):287-301. PMC: 5990995. DOI: 10.1007/s13244-018-0599-0. View

10.

Jacobs M, Pradier M, McCoy Jr T, Perlis R, Doshi-Velez F, Gajos K . How machine-learning recommendations influence clinician treatment selections: the example of the antidepressant selection. Transl Psychiatry. 2021; 11(1):108. PMC: 7862671. DOI: 10.1038/s41398-021-01224-x. View

11.

van Leeuwen K, de Rooij M, Schalekamp S, van Ginneken B, Rutten M . How does artificial intelligence in radiology improve efficiency and health outcomes?. Pediatr Radiol. 2021; 52(11):2087-2093. PMC: 9537124. DOI: 10.1007/s00247-021-05114-8. View

12.

Hoff T . Professional commitment among US physician executives in managed care. Soc Sci Med. 2000; 50(10):1433-44. DOI: 10.1016/s0277-9536(99)00410-4. View

13.

Lin M, Pappas S, Sellin J, El-Serag H . Curbside Consultations: The Good, the Bad, and the Ugly. Clin Gastroenterol Hepatol. 2015; 14(1):2-4. DOI: 10.1016/j.cgh.2015.09.026. View

14.

Ghassemi M, Oakden-Rayner L, Beam A . The false hope of current approaches to explainable artificial intelligence in health care. Lancet Digit Health. 2021; 3(11):e745-e750. DOI: 10.1016/S2589-7500(21)00208-9. View

15.

Johnson A, Pollard T, Berkowitz S, Greenbaum N, Lungren M, Deng C . MIMIC-CXR, a de-identified publicly available database of chest radiographs with free-text reports. Sci Data. 2019; 6(1):317. PMC: 6908718. DOI: 10.1038/s41597-019-0322-0. View

16.

Dietvorst B, Simmons J, Massey C . Algorithm aversion: people erroneously avoid algorithms after seeing them err. J Exp Psychol Gen. 2014; 144(1):114-26. DOI: 10.1037/xge0000033. View

17.

Gaube S, Suresh H, Raue M, Merritt A, Berkowitz S, Lermer E . Do as AI say: susceptibility in deployment of clinical decision-aids. NPJ Digit Med. 2021; 4(1):31. PMC: 7896064. DOI: 10.1038/s41746-021-00385-9. View

18.

Hekman D, Steensma H, Bigley G, Hereford J . Effects of organizational and professional identification on the relationship between administrators' social influence and professional employees' adoption of new work behavior. J Appl Psychol. 2009; 94(5):1325-35. DOI: 10.1037/a0015315. View