Table 1. Receiver operating characteristic curve statistics and cutoff thresholds for transition predictors.

Variable group	Variables	n	Youden index	Score threshold	AUROC	P
Historical	Gender	40	0.05329	1	0.527	0.7613
	Age	40	0.2414	⩽17.2841	0.517	0.8563
	Family history of psychosis	39	0.06207	1	0.531	0.6758
	Duration of symptoms at presentation	40	0.2571	⩽7	0.57	0.5125
	Any history of drug usea	40	0.4514	1	0.726	0.006b
Clinical assessment	PANSS positive	40	0.4734	>14	0.699	0.0226b
	PANSS negative	40	0.4169	>12	0.723	0.027b
	PANSS general	40	0.4859	>31	0.768	0.0043b
	MADRS	40	0.3448	>12	0.666	0.0664
	GAF—low	40	0.4263	⩽60	0.774	0.0002b
Blood biomarkers	Nervonic acid	39	0.5357	⩽0.2902	0.688	0.0235b
	PLA2	39	0.276	⩽0.6473	0.562	0.6056
	Total Omega 3	39	0.4968	⩽5.0727	0.724	0.0405b
	SOD	36	0.24	>0.0318	0.502	0.9859
	GSH	36	0.3855	⩽40.024	0.691	0.0522
	GSSG	36	0.1455	⩽16.25	0.527	0.7958
	GSH/GSSG	36	0.2036	⩽1.877	0.571	0.5142
qEEG measures	Delta frontal	34	0.2933	⩽1.82	0.542	0.7482
	Alpha frontal	34	0.1733	⩽1.71	0.507	0.9573
	Beta1 occipitoparietal	34	0.2800	⩽1.6	0.636	0.2111

Abbreviations: AUROC, area under the receiver operating characteristic curve; GAF, Global assessment of functioning; GSH, glutathione; GSSG, glutathione disulfide (oxidized form); MADRS, Montgomery Asberg Depression Rating Scale; PANSS, Positive And Negative Symptom Scale; PLA2, Phospholipase A2; qEEG, resting quantitative electroencephalography; SOD, superoxide dismutatse.

For continuous variables cutoff thresholds were set at the optimum balance between sensitivity and specificity as determined by calculation of the Youden index. The threshold value for dichotomous variables was 1.

^aAny history of drug use=illicit, tobacco or alcohol.

^bSignificant predictors of transition at P<0.05 are in bold.