Skip to main content
. Author manuscript; available in PMC: 2020 Sep 28.
Published in final edited form as: Proc Mach Learn Res. 2020 Aug;126:479–507.

Table 3:

Results of different balancing approaches and their influence on the performance on Duke dataset (average AUC on leave one patient out cross-validation). Standard deviation (SD) is shown in parentheses. Results do not include post-processing routine.

Model No Balancing Class Balancing Patient-Class Balancing
DL1 0.933 (0.055) 0.923 (0.070) 0.911 (0.086)
DL2 0.923 (0.057) 0.950 (0.041) 0.943 (0.051)
SVM 0.822 (0.063) 0.772 (0.061) 0.765 (0.058)