Skip to main content
. 2019 Nov 1;10:1105. doi: 10.3389/fneur.2019.01105

Table 3.

Training results.

Atlas Discover Accuracy P value Precision Sensitivity Specificity 2nd test
Feature selection Brod. 0.0711 0.900 <0.001 0.8889 0.8889 0.9091 0.6667
Brod.ce 0.0739 0.900 <0.001 0.8889 0.8889 0.9091 0.5556
AAL 0.0699 0.925 <0.001 0.9412 0.8889 0.9545 0.7222
AICHA 0.0669 1.000 <0.001 1.000 1.000 1.000 0.8333
BN 0.0765 1.000 <0.001 1.000 1.000 1.000 0.8333
Whole brain Brod. 0.5250 0.3816 0.4667 0.3889 0.6364
Brod.ce 0.5000 0.4645 0.4375 0.3889 0.5909
AAL 0.4750 0.5455 0.4211 0.4444 0.5000
AICHA 0.6000 0.1269 0.5833 0.3889 0.7727
BN 0.5750 0.1888 0.5385 0.3889 0.7273

The training results were divided into with or without feature selection. The Discover column shows the ratio of significant connections to all possible connections. The last column shows the test accuracy when using the second session data as test data. This testing was only performed for the training with feature selection. Brod, Brodmann atlas; Brod.ce, Brodmann atlas with cerebellum; BN, Brainnetome atlas. Significant results were shown in bold.