Skip to main content
. 2022 Mar 11;13:864393. doi: 10.3389/fpsyt.2022.864393

Table 3.

Comprehensive comparison of existing state-of-the-art methods with proposed method.

Method Subject Channel Feature Protocol Accuracy
(MDD,HC) (%)
Feature-level fusion (26) (86, 92) EEG (3) 60 linear and Ten-fold CV 86.98
36 nonlinear features
Multivariate pattern analysis (27) (27, 28) EEG (128) 249 EEG features LOSOCV 92.73
Multimodal fusion (28) (81,89) EEG (3) 6 EEG features Nested CV 86.64
and voice(1) and 15 voice features
Case-Based Reasoning Model (29) (86, 92) EEG (3) 113 EEG features Ten-fold CV 91.25
SVM (30) (20, 19) EEG (64) 3 potential biomarker Ten-fold CV 89.7
KNN (31) (92, 121) EEG (3) 270 features Ten-fold CV 79.27
Independent component analysis (32) (13, 13) EEG (64) - - -
Brain Function Networks (33) (24, 24) EEG (64) LC-CC in theta band Ten-fold CV 93.31
Correlated Feature Selection (23) (17, 17) EEG (128) 10 EEG features LOSOCV 88.94
Ours (24, 29) EEG (128) - LOSOCV 90.98

CV, Cross-Validation; LOSOC, Leave-One-Subject-Out Cross-Validation;MDD, Major Depression Disorder; HC, Healthy Control.