Skip to main content

Table 1 (abstract P20). Summary of model performance

From: Methods for Evaluation of medical prediction Models, Tests And Biomarkers (MEMTAB) 2018 Symposium

Model Discrimination (c-statistic) Calibration in the large (alpha) Logistic miscalibration (beta)
CE-MARC: n=675, Mean age 59.4, 61.8% male, Median CAD PTL by [Duke Clinical/CAD Consortium Basic/Clinical] = [56.2%, 18.3%, 17.6%]. CAD detected 235/650 (36.2%)
 CE-MARC (2016) 0.779 (0.742, 0.814) NA NA
 Duke Clinical Risk Score (1993) 0.763 (0.725, 0.800) -1.108 (-1.305, 0.911); P<0.001 -0.298 (-0.416, -0.180); P<0.001
 CAD Consortium (2012) Basic 0.770 (0.733, 0.806) 0.713 (0.532, 0.893); P<0.001 -0.015 (-0.131, 0.101); P=0.803
 CAD Consortium (2012) Clinical 0.762 (0.725, 0.7995) 0.822 (0.639, 1.005); P<0.001 -0.051 (-0.159, 0.057); P=0.354
CE-MARC2+ (n=369, Mean age 59.6, 43.4% male, Median CAD PTL by [Duke Clinical/CAD Consortium Basic/Clinical] = [67.4%, 16.8%, 14.8%]. CAD detected 158/369 (42.8%)
 CE-MARC (2016) 0.777 (0.731, 0.824) 0.045 (-0.190, 0.280); P=0.709 0.028 (-0.214, 0.269); P=0.823
 Duke Clinical Risk Score (1993) 0.752 (0.704, 0.801) -1.016 (-1.265, -0.766); P<0.001 -0.207 (-0.363, -0.050); P=0.010
 CAD Consortium (2012) Basic Model 0.755 (0.706, 0.803) 0.738 (0.507, 0.969); P<0.001 -0.007 (-0.182, 0.169); P=0.940
 CAD Consortium (2012) Clinical 0.752 (0.703, 0.800) 0.866 (0.629, 1.103); P<0.001 -0.054 (-0.121, 0.105); P=0.507