Table 2.

Receiver operating characteristic (ROC) curve analyses on supportive vector machine (SVM) classifiers of Weibo users’ suicide probability and emotional distress.

Outcome Variables	n (%)	AUCa (95% CI)	P value	Sensitivityb (%)	Specificity (%)
For all respondents (N=976)
Had Weibo suicide communication	117 (12)	0.56 (0.50-0.61)	.06	61	49
High suicide risk (SPS≥80)	190 (19)	0.48 (0.44-0.53)	.43	64	32
Severe depression (DASSc-Depression score>10)	49 (5)	0.47 (0.38-0.55)	.41	63	33
Severe anxiety (DASS-Anxiety score>7)	140 (14)	0.45 (0.40-0.50)	.06	58	32
Severe stress (DASS - Stress score > 12)	45 (5)	0.47 (0.39-0.56)	.52	64	33
For respondents who had Weibo suicide communication (N=117)
High suicide risk (SPSd≥80)	51 (44)	0.61 (0.51-0.72)	.04	65	58
Severe depression (DASS-Depression score>10)	23 (20)	0.57 (0.42-0.72)	.31	65	50
Severe anxiety (DASS-Anxiety score>7)	43 (37)	0.75 (0.65-0.84)	<.001	70	66
Severe stress (DASS-Stress score>12)	20 (17)	0.64 (0.52-0.76)	.05	65	57

^aAUC: area under the curve.

^bSensitivity and specificity at the optimal cut-off were reported.

^cDASS: Depression Anxiety Stress Scales.

^dSPS: Suicide Probability Scale.