Table 2.

Performance of SVM classifiers using only PSD, only PAC, or both PSD and PAC features

		Sensitivity/Specificity
	Error	Naive	Mild	Moderate	Severe
	PSD (134 features)
P1	0.11	0.96/0.94	0.83/0.95	0.81/0.97	0.92/0.97
P2	0.14	0.94/0.88	0.84/0.99	0.70/0.95	0.85/0.96
P3	0.14	0.89/0.94	0.69/0.97	0.91/0.84
All	0.19	0.90/0.85	0.76/0.96	0.73/0.91	0.79/0.98
	PAC (72 features)
P1	0.52	0.77/0.40	0.26/0.74	0.10/0.89	0.47/0.85
P2	0.59	0.72/0.16	0.11/0.78	0.17/0.84	0.12/0.87
P3	0.57	0.35/0.59	0.08/0.88	0.61/0.31
All	0.61	0.66/0.29	0.16/0.74	0.19/0.69	0.27/0.90
	PSD + PAC (206 features)
P1	0.19	0.96/0.82	0.76/0.93	0.66/0.97	0.71/0.97
P2	0.21	0.92/0.73	0.73/0.97	0.53/0.95	0.76/0.96
P3	0.22	0.79/0.86	0.59/0.99	0.85/0.76
All	0.28	0.87/0.74	0.67/0.92	0.58/0.89	0.62/0.96

SVM, support vector machine; PSD, power spectral density; PAC, phase-amplitude coupling.