Table 3.

Stratification analysis of HMDT on training and validation cohorts.

Subpopulation	Training cohort (n = 204)				Validation cohort (n = 88)
	AUC (95% CI)	ACC	SEN	SPE	AUC (95% CI)	ACC	SEN	SPE
Age
<44	0.971 (0.921, 1)	0.941	0.946	0.933	0.894 (0.797, 0.990)	0.829	0.846	0.800
≥44	0.991 (0.948, 1)	0.966	0.982	0.954	0.933 (0.865, 1)	0.894	0.900	0.889
Tumor shape
Yes	1	0.950	1	0.889	0.930 (0.805, 1)	0.667	1	0.500
No	0.983 (0.963,1)	0.962	0.980	0.941	0.924 (0.866,0.982)	0.863	0.829	0.906
Dural tail sign
Yes	0.983 (0.957, 1)	0.929	1	0.904	0.944 (0.849, 1)	0.739	1	0.667
No	0.978 (0.945, 1)	0.970	0.989	0.930	0.904 (0.931, 0.976)	0.831	0.878	0.750
WHO grade I meningiomas	0.988 (0.968, 1)	0.968	0.973	0.961	0.914 (0.854, 0.973)	0.854	0.848	0.861
Angiomatous meningiomas	0.997 (0.992, 1)	0.978	0.991	0.920	0.913 (0.816, 1)	0.873	0.935	0.556

AUC, area under the curve; CI, confidence interval; ACC, accuracy; SEN, sensitivity; SPE, specificity.