Skip to main content

Table 2 Average train and test AUC performance of models trained in the double loop cross validation pipeline

From: A machine learning approach using endpoint adjudication committee labels for the identification of sepsis predictors at the emergency department

 

Demographic and vital

(P = 9)

Laboratory

(P = 10)

Advanced haematology

(P = 51)

All

(P = 70)

 

Train

Test

Train

Test

Train

Test

Train

Test

Logistic

regression

0.87

(0.86–0.89)

0.84

(0.79–0.89)

0.78

(0.76–0.80)

0.71

(0.64–0.77)

0.90

(0.89–0.91)

0.66

(0.59–0.73)

0.99

(0.99–0.99)

0.73

(0.66–0.81)

Lasso

0.87

(0.85–0.88)

0.84

(0.79–0.89)

0.77

(0.75–0.79)

0.71

(0.64–0.80)

0.82

(0.82–0.85)

0.77

(0.71–0.83)

0.94

(0.93–0.95)

0.85

(0.80–0.95)

Random

Forest

0.90

(0.88–0.91)

0.83

(0.76–0.90)

0.88

(0.86–0.89)

0.69

(0.62–0.77)

0.91

(0.90–0.92)

0.76

(0.69–0.83)

0.96

(0.95–0.97)

0.84

(0.78–0.89)

  1. P denotes the number of variables in each model set. 95% confidence intervals are shown in parentheses for each model and data configuration