Skip to main content

Table 5 Model evaluation

From: A practical method for predicting frequent use of emergency department care using routinely available electronic registration data

   Multivariable Logistic Regression Models
Model No. 1 2 3 4 5 6 7 8 9
No. of visits constituting ‘frequent use’ > = 8 > = 9 > = 10 > = 11 > = 12 > = 13 > = 14 > = 15 > = 16
Area under ROC curve (AUC) 0.84 0.85 0.87 0.88 0.89 0.89 0.90 0.91 0.92
With sensitivity < =25 %, probability > 0.5
PPV (%) 64.5 63.9 63.4 62.9 62.4 61.3 60.8 60.6 59.5
Specificity (%) 99.5 99.6 99.7 99.7 99.8 99.8 99.8 99.8 99.9
False positive patients Total No. 5883 4923 4125 3447 2974 2610 2273 2007 1805
> = 8 visits (No.) 0 565 843 1021 1071 1103 1077 1038 998
Adjusted PPV for patients with > =8 visits in subsequent two years (%) 64.5 68 70.9 73.9 75.9 77.7 79.4 81 81.9