Skip to main content
. 2020 Dec 31;3(4):602–610. doi: 10.1093/jamiaopen/ooaa058

Table 2.

Overall AUROC, AUPRC, and ECE for LR and HLR models over all 73 target variables and across all physicians

Measure LR HLR P-value
AUROC 0.75 (0.74–0.76) 0.81 (0.800.82) <.001
AUPRC 0.665 (0.663–0.667) 0.763 (0.7620.765) <.001
ECE 0.16 (0.14–0.17) 0.07 (0.060.08) <.001

Higher AUROC and AUPRC show better discrimination power while lower ECE denotes better probability calibration. The best values for each metric are in boldface.