highly significantp = 1.39 × 10 −17
The changes in AUC-ROC were not statistically significant (DeLong p -values of 0.922, 0.902, and 0.750 respectively); however, the McNemar test was highly significant ( p = 1.39 × 10 −17 , 5.23 × 10 −30 , 2.25 × 10 −24 ).