Skip to main content

Table 3 HSI trained on all features is evaluated under specific operating modes where limited set of features are accessible to the model

From: Early prediction of hemodynamic interventions in the intensive care unit using machine learning

Operating mode

AUC

Sp (BE)

PPV (BE)

Se (Sp = 90%)

PPV (Sp = 90%)

Se (Sp = 95%)

PPV (Sp = 95%)

All features

0.82

0.92

0.52

0.55

0.49

0.43

0.6

Basic

0.72

0.89

0.39

0.37

0.4

0.26

0.48

Basic + labs

0.8

0.91

0.48

0.5

0.47

0.37

0.57

Basic + ventilation

0.76

0.9

0.45

0.46

0.45

0.35

0.55

  1. The outcome label includes pressors, fluid, and PRBC administration. We report the area under the receiver operator curve (AUC), Specificity (Sp), Sensitivity (Se) and Positive Predictive Value (PPV) at the breakeven point (BE) where precision equals recall, and at both 90% and 95% Specificity
Back to article page

Critical Care

ISSN: 1364-8535