Skip to main content
. 2023 Nov 1;4(11):101257. doi: 10.1016/j.xcrm.2023.101257

Table 2.

Diagnostic performance of C-DAG algorithm for identifying patients with NASH

Training set (n = 326) Test set (n = 163) External validation cohort (n = 135) Pooled validation cohort (n = 298)
Number of patients with NASH 125 66 43 109
AUROC (95% CI) 0.911 (0.881–0.941) 0.891 (0.844–0.939) 0.895 (0.844–0.946) 0.893 (0.859–0.927)
Delong’s test p (vs. sCDCP1) <0.001 0.071 0.017 0.004
Rule-out cutoff <0.235 <0.235 <0.235 <0.235
n (%) 166 (50.9) 69 (42.3) 77 (57.0) 145 (48.7)
Sensitivity, % 90.3 93.9 86.0 90.8
Specificity, % 76.1 66.0 77.2 71.4
NPV, % 92.7 94.1 92.2 93.1
Rule-in cutoff ≥0.393 ≥0.393 ≥0.393 ≥0.393
n (%) 111 (34.0) 73 (44.8) 42 (31.1) 115 (38.6)
Sensitivity, % 72.8 77.3 65.1 72.5
Specificity, % 90.0 78.4 85.9 81.5
PPV, % 82.0 70.8 68.3 69.3
Gray zone n (%) 49 (15.0) 21 (12.9) 16 (11.9) 38 (12.8)

The pooled validation cohort combines the test sets of the main cohort and the external validation cohort. NASH, non-alcoholic steatohepatitis; PPV, positive predictive value; NPV, negative predictive value.