CT-derived fractional flow reserve for prediction of major adverse cardiovascular events in diabetic patients

Table 4 Prediction accuracy and risk reclassification of each model

Models	P Value by the Log-likelihood test	C-index (95%CI)	P Value	Time-dependent AUC		AIC	Brier score
Models	P Value by the Log-likelihood test	C-index (95%CI)	P Value	1-year	3-year	AIC	1-year	3-year
Training set
Model 1	< 0.001	0.80 (0.75–0.85)	0.021	0.82	0.82	801.08	0.011	0.026
Model 2		0.82 (0.77–0.87)		0.85	0.84	786.25	0.011	0.024
Internal validation set
Model 1	< 0.001	0.80 (0.73–0.87)	0.022	0.80	0.84	384.64	0.025	0.039
Model 2		0.84 (0.77–0.90)		0.83	0.87	371.56	0.021	0.035
External test set
Model 1	< 0.001	0.82(0.74–0.90)	0.027	0.89	0.82	328.27	0.017	0.031
Model 2		0.85(0.77–0.93)		0.92	0.87	313.06	0.015	0.025

AIC, Akaike information criterion; C-index, Harrell’s C-statistics
Model 1 included age, sex, BMI, hypertension, diabetes, dyslipidemia, current-smoking, fast glucose, HbA1c, Obstructive CAD, CACS, HRP, LAP, PR, SC and NRS
Model 2 = Model 1 + CT-FFR

ISSN: 1475-2840