Skip to main content

Table 3 Performance of rule-based model vs. baseline models

From: Understanding the complexity of sepsis mortality prediction via rule discovery and analysis: a pilot study

Method

Mean AUC (SD)

P value

SAPS-II

0.794 (0.017)

\(<0.01\)

LODS

0.757 (0.016)

\(<0.01\)

SOFA

0.743 (0.020)

\(<0.01\)

qSOFA

0.56 (0.018)

\(<0.01\)

SIRS

0.564 (0.018)

\(<0.01\)

SVM

0.675 (0.015)

\(<0.01\)

Random Forest

0.687 (0.017)

\(<0.01\)

LASSO

0.766 (0.018)

\(<0.01\)

Ridge regression

0.765 (0.017)

\(<0.01\)

Logistic regression

0.765 (0.018)

\(<0.01\)

Rule-based

0.781 (0.018)

-

  1. The third column (p value) shows the significance of difference in AUCs (obtained via the Delong test) of each model against the rule-based model. The largest three AUC values are shown in bold