TABLE 5.

Accuracy of different models for predicting Parkinson’s disease with major depression.

Model	AUC (95%CI)	Sensitivity	Specificity	P
Basic Age Disease duration UPDRS III score H&Y stage Subtype of PD	0.847 (0.744–0.950)	68.8%	61.1%	<0.001
Basic+ Cortical thickness of seed 1 Cortical thickness of seed 2	0.874 (0.779–0.969)	78.1%	66.6%	<0.001
Basic+ FC between seed 1 and right cerebellum posterior lobe FC between seed 2 and right inferior frontal gyrus and insula	0.904 (0.819–0.989)	87.5%	72.1%	<0.001
Full model*	0.927 (0.854–0.999)	93.8%	78.4%	<0.001

AUC, area under the receiver operating characteristic curve; CI, confidence interval; FC, functional connectivity; H&Y, Hoehn and Yahr scale; UPDRS III, Unified PD Rating Scale part III.

*Full model, combining basic clinical features (age, disease duration, UPDRS III score, H&Y stage, and subtype of PD) with the following neuroimaging biomarkers: cortical thickness of the two seeds and the FC values between seed 1 and the right cerebellum posterior lobe as well as between seed 2 and the right inferior frontal gyrus and insula.