Artificial neural networks improve early outcome prediction and risk classification in out-of-hospital cardiac arrest patients admitted to intensive care

Table 2 Predictor ranking and prediction performance in data set A

The ranking of the variables in the data set was calculated by their individual effect on the AUC when subtracted from the overall model. The AUC values in Table 2a represent how much the performance decreases when the corresponding variable is excluded from the model
The AUC values in Table 2b show how much the prediction performance of the model increase by adding one variable at the time (based on their relative importance in Table 2a) to the model. AUC_CV values represent the training set, and AUC_test the test set. When using all available variables at patient admission to intensive care, the AUC was 0.891, indicating an excellent performance of predicting long-term functional outcome
AUC area under the curve, ROSC return of spontaneous circulation, GCS Glasgow Coma Scale, AMI acute myocardial infarction, CPR cardiopulmonary resuscitation

ISSN: 1364-8535