Skip to main content

Table 3 Performance of global and clustering models (GLM model: generalized linear models with variables according ICU mortality table; Global model: GLM model with important variables according to information value analysis for ICU mortality; A, B and C Phenotypes: GLM models with important variables according to information value analysis for each cluster)

From: Deploying unsupervised clustering analysis to derive clinical phenotypes and risk factors associated with mortality risk in 2022 critically ill patients with COVID-19 in Spain

Variable

Classic model

Automatic model

A Phenotype

B phenotype

C phenotype

No. patients

2022

2017

537

623

857

Variables included

25

25

26

19

18

Accuracy

0.78

0.77

0.86

0.80

0.72

Sensitivity

0.88

0.88

0.94

0.92

0.71

Specificity

0.45

0.54

0.55

0.47

0.73

AUC ROC

0.83

0.83

0.90

0.85

0.79