Table 2.

Diagnostic performance

Method and threshold	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Accuracy (%)	AUC
CCTA ≥ 25% DS	100.0 (81/81) [95.5–100.0]	22.2 (10/45) [11.2–37.1]	69.8 (81/116) [66.4–73.0]	100.0 (10/10) [100.0–100.0]	72.2 (91/126) [64.3–80.2]	0.68 [0.59–0.76]
CCTA ≥ 50% DS	92.6 (75/81) [84.6–97.2]	31.1 (14/45) [18.2–46.6]	70.8 (75/106) [66.3–74.8]	70.0 (14/20) [49.1–85.0]	70.6 (89/126) [62.6–78.7]	0.68 [0.59–0.76]
CCTA ≥ 70% DS	17.3 (14/81) [9.8–27.3]	97.8 (44/45) [88.2–99.9]	93.3 (14/15) [65.5–99.0]	39.6 (44/111) [37.1–42.3]	46.0 (58/126) [37.2–54.9]	0.68 [0.59–0.76]
CCTA DS + DL combined	84.6 ± 0.03	48.4 ± 0.04	74.7 ± 0.01	64.0 ± 0.05	71.7 ± 0.02	0.76 ± 0.02

Data is given in percentage, data in parentheses is raw data, and data in square brackets is 95% confidence interval. For the combined method, data is depicted as average ± SD of 50 cross-validation experiments. AUC area under the receiver operating characteristic curve, CCTA coronary computed tomography angiography, DL deep learning, DS degree of stenosis, NPV negative predictive value, PPV positive predictive value, SD standard deviation