Who Goes First? Influences of Human-AI Workflow on Decision Making in Clinical Imaging.

Fogliato R, Chappidi S, Lungren M, Fisher P, Wilson D, Fitzke M, et al. Who Goes First? Influences of Human-AI Workflow on Decision Making in Clinical Imaging. FAccT '22: Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency. 2022; Pages 1362-1374. https://doi.org/10.1145/3531146.3533193

Summary: This study examines how the sequencing of AI assistance affects decision-making in clinical imaging, focusing on radiology workflows. A user study with 19 veterinary radiologists explored two configurations: AI inferences presented before or after radiologists made provisional diagnoses. The findings revealed that radiologists were less likely to agree with AI when their provisional decision preceded AI review, and they were less inclined to seek second opinions when disagreeing with AI. Surprisingly, this configuration did not increase time spent on tasks. The research provides insights into designing effective human-AI collaboration in clinical settings and offers an open-source platform for future studies.