Skip to main content
. 2021 Mar 16;11:6006. doi: 10.1038/s41598-021-85570-2

Figure 3.

Figure 3

(a) AUPR performance on the test set #1 for DeepWrist, radiology resident, two radiologists (R1 & R2), and two primary care physicians (PCP1 & PCP2), (b) AUPR performance of DeepWrist and other graders on the test set #2. The plot highlights the drop in performance for both—human raters of different expertise, and our method.