Skip to main content

Table 2 Model testing (Elixhauser comorbidities model)

From: Machine learning-derived prediction of in-hospital mortality in patients with severe acute respiratory infection: analysis of claims data from the German-wide Helios hospital network

Algorithm

AUC (95%CI)

AUPRC (95%CI)

GLM

0.83 (0.825–0.834)

0.372 (0.361–0.384)

RF

0.831 (0.827–0.835)

0.384 (0.373–0.396)

NNET

0.834 (0.83–0.838)

0.382 (0.371–0.393)

XGBoost

0.834 (0.83–0.839)

0.389 (0.378–0.4)

  1. 95% CI 95% confidence interval, AUC Area under the curve, AUPRC Area under the precision-recall curve, GLM generalized linear models, NNET single layer neural network, RF random forest, XGBoost extreme gradient boosting