Table 3 The performance of five models in the training and testing cohort (15 mmHg).

Signature	Accuracy	AUC	95% CI	Sensitivity	Specificity	PPV	NPV	Cohort
Clinic
	0.775	0.868	0.796–0.925	0.756	0.784	0.62	0.873	Train
	0.750	0.801	0.689–0.913	0.640	0.821	0.696	0.780	Test
Rad
	0.915	0.974	0.952–0.995	0.951	0.898	0.812	0.975	Train
	0.891	0.912	0.832–0.991	0.800	0.949	0.909	0.881	Test
DTL
	0.881	0.946	0.908–0.984	0.878	0.841	0.719	0.936	Train
	0.783	0.850	0.747–0.956	0.759	0.897	0.826	0.854	Test
DLR
	0.884	0.973	0.951–0.994	0.951	0.852	0.75	0.974	Train
	0.906	0.914	0.835–0.994	0.840	0.949	0.913	0.902	Test
Nomogram
	0.907	0.974	0.953–0.995	0.951	0.886	0.796	0.975	Train
	0.891	0.919	0.846–0.993	0.84	0.923	0.875	0.9	Test

DTL, deep transfer learning; AUC, area under curve; CI, confidence interval. DLR, Deep Learning Radiomics.

Quick links

Search