Assessing Predictive Ability of Dynamic Time Warping Functional Connectivity for ASD Classification

Table 4

Results from single iterations of nested LOOCV and optimistically biased CV.

Method	Accuracy	Sensitivity	Specificity
Nested LOOCV (DTW)
SVM-Linear	0.58	0.51	0.64
SVM-Linear (PCA)	0.71	0.69	0.72
SVM-Radial	0.56	0.53	0.58
SVM-Radial (PCA)	0.67	0.67	0.66
L1-SVM	0.59	0.57	0.60
L1-SVM (PCA)	0.66	0.63	0.68
Nested LOOCV (PC)
SVM-Linear	0.62	0.61	0.62
SVM-Linear (PCA)	0.62	0.61	0.62
SVM-Radial	0.54	0.51	0.56
SVM-Radial (PCA)	0.53	0.47	0.58
L1-SVM	0.59	0.53	0.64
L1-SVM (PCA)	0.59	0.61	0.56
Optimistically biased CV (DTW)
SVM-Linear	0.60	0.57	0.62
SVM-Linear (PCA)	0.67	0.70	0.64
SVM-Radial	0.61	0.47	0.74
SVM-Radial (PCA)	0.67	0.61	0.72
L1-SVM	0.66	0.68	0.64
L1-SVM (PCA)	0.66	0.56	0.76
Optimistically biased CV (PC)
SVM-Linear	0.60	0.54	0.66
SVM-Linear (PCA)	0.58	0.56	0.60
SVM-Radial	0.60	0.55	0.64
SVM-Radial (PCA)	0.63	0.48	0.78
L1-SVM	0.66	0.68	0.64
L1-SVM (PCA)	0.63	0.66	0.60