Skip to main content

Table 3 Calibration metrics

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

 

Calibration-in-the-large

Calibration intercept (95%CI)

Calibration slope (95%CI)

GLM

11.5% (6969/60414) vs. 11.7%

− 0.02 (− 0.049 to 0.006)

1.02 (0.997 to 1.05)

RF

11.5% (6969/60414) vs. 11.8%

− 0.03 (− 0.059 to − 0.005)

1.26 (1.228 to 1.299)

NNET

11.5% (6969/60414) vs. 11.7%

− 0.02 (− 0.05 to 0.005)

1.01 (0.982 to 1.038)

XGBoost

11.5% (6969/60414) vs. 11.7%

− 0.02 (− 0.051 to 0.004)

1.03 (1.003 to 1.057)

  1. 95% CI 95% confidence interval, GLM generalized linear models, NNET single layer neural network, RF random forest, XGBoost extreme gradient boosting
Back to article page

Respiratory Research

ISSN: 1465-993X

Contact us