. 2022 Feb 17;12:2755. doi: 10.1038/s41598-022-06651-4

Table 3.

Classification performance using all measures.

Study and description	Classifier(s)	Parameters tuned	Accuracy	Sensitivity	Specificity	F1	AUC
6—All subcortical and cortical measures	Logistic Regression	Regularization parameter C, Max iterations, penalty, solver	77%	79%	74%	0.83	0.77
7—Ensemble with 3 inputs (a) SV (b) CV (c) CA + CT + CMC	Ensemble—Hard Voting (a) Support Vector Classifier (b) Nu-Support Vector Classifier (c) Logistic Regression		83%	90%	70%	0.83	0.80
8—Ensemble with 5 inputs (a) SV (b) CV (c) CA (d) CT € CMC	Ensemble—Soft Stacking (a) Support Vector Classifier (b) Nu-Support Vector Classifier (c) Support Vector Classifier (d) Support Vector Classifier (e) Logistic Regression	Regularization parameter C, Max iterations, penalty, solver	87%	98%	65%	0.87	0.82

Study and description

Classifier(s)

Parameters tuned

Accuracy

Sensitivity

Specificity

AUC

6—All subcortical and cortical measures

Logistic Regression

Regularization parameter C, Max iterations, penalty, solver

77%

79%

74%

0.83

0.77

7—Ensemble with 3 inputs

(a) SV

(b) CV

Ensemble—Hard Voting

(a) Support Vector Classifier

(b) Nu-Support Vector Classifier

83%

90%

70%

0.83

0.80

8—Ensemble with 5 inputs

(a) SV

(b) CV

(d) CT

€ CMC

Ensemble—Soft Stacking

(a) Support Vector Classifier

(b) Nu-Support Vector Classifier

(d) Support Vector Classifier

(e) Logistic Regression

Regularization parameter C, Max iterations, penalty, solver

87%

98%

65%

0.87

0.82