Table 4 Sensitivity, specificity, accuracy, F-measure and area under the curve of artificial intelligence models
Model type | Name of the model | Sensitivity | Specificity | Accuracy | F-measure | Area under the ROC curve |
|---|---|---|---|---|---|---|
Conventional | SOFA | 66.67 | 71.40 | 69.52 | 63.43 | 76.0 |
SAPS II | 67.26 | 73.37 | 70.95 | 64.73 | 77.1 | |
APACHE II | 73.9 | 73.0 | 73.3 | 68.71 | 80.3 | |
APACHE IV | 73.6 | 69.4 | 71.1 | 66.86 | 78.5 | |
Artificial intelligence | MLP NNsel | 84.68 | 72.26 | 77.82 | 77.36 | 84.1 |
CART DTsel | 80.18 | 72.99 | 76.21 | 75.10 | 80.0 | |
MLP NNall | 81.16 | 75.44 | 77.91 | 76.34 | 83.2 | |
CART DTall | 79.81 | 72.34 | 75.6 | 74.03 | 77.7 | |
Artificial intelligence (external validation) | MLP NNsel | 66.67 | 83.70 | 81.69 | 46.15 | 78.9 |
CART DTsel | 83.33 | 51.85 | 55.55 | 30.61 | 72.9 | |
MLP NNall | 83.3 | 77.8 | 78.43 | 67.63 | 82.3 | |
CART DTall | 71.1 | 82.22 | 79.73 | 59.37 | 75.6 |