Detecting Depression Using an Ensemble Logistic Regression Model Based on Multiple Speech Features

Table 5

Classification outcomes of each individual classifier for females.

Features	SVM			GMM			LR
	Sen. (%)	Spe. (%)	Acc. (%)	Sen. (%)	Spe. (%)	Acc. (%)	Sen. (%)	Spe. (%)	Acc. (%)
MFCC	63.24	57.27	60.31	56.47	66.06	61.17	62.79	61.80	62.30
PROS	67.21	60.65	63.99	51.72	73.29	62.30	64.35	66.73	65.52
SPEC	60.64	63.35	61.97	52.44	73.70	62.87	63.05	64.91	63.96
GLOT	56.60	42.53	49.70	51.33	50.44	50.90	52.70	46.11	49.47
MFCC + PROS	67.53	61.06	64.36	56.86	71.54	64.06	64.93	66.06	65.48
MFCC + SPEC	66.10	61.60	63.89	57.78	69.78	63.66	63.24	65.99	64.59
MFCC + GLOT	63.05	57.20	60.18	55.63	64.84	60.15	62.66	60.24	61.47
PROS + SPEC	64.09	64.50	64.29	51.01	73.83	62.20	63.63	67.61	65.58
PROS + GLOT	67.47	59.16	63.40	52.31	72.08	62.00	63.37	66.73	65.02
SPEC + GLOT	61.09	60.31	60.71	51.79	70.99	61.21	61.87	62.75	62.30
MFCC + PROS + SPEC	64.74	62.41	63.59	56.47	73.09	64.62	64.41	67.41	65.88
MFCC + PROS + GLOT	67.08	62.27	64.72	55.50	72.62	63.89	64.74	67.14	65.92
MFCC + SPEC + GLOT	63.37	62.68	63.03	57.71	69.37	63.43	62.39	63.42	62.90
PROS + SPEC + GLOT	64.15	63.42	63.79	51.53	73.43	62.27	63.11	67.07	65.05
MFCC + PROS + SPEC + GLOT	65.00	63.22	64.13	56.02	72.96	64.32	63.44	67.61	65.48

Maximum of sensitivity (sen.), specificity (spe.), and accuracy (acc.) are shown in bold.