Skip to main content

Table 2 Performance of elastic net and neural networks to predict ICU mortality (AUC ± SE).

From: Updating mortality risk estimation in intensive care units from high-dimensional electronic health records with incomplete data

  

Complete-case

n = 4276

  

Imputed-19

n = 10,089

 

Imputed-70

n = 16,532

Admission data

6-hour slots

12-hour slots

 

12-hour slots

 

12-hour slots

Elastic net

0.709 ± 0.002

0.769 ± 0.029

0.785 ± 0.002

 

0.753 ± 0.024

 

0.777 ± 0.003

FCN

0.663 ± 0.055

0.521 ± 0.037

0.634 ± 0.049

 

0.586 ± 0.056

 

0.542 ± 0.047

CNN

0.742 ± 0.002

0.778 ± 0.005

0.778 ± 0.006

 

0.751 ± 0.022

 

0.783 ± 0.003

LSTM

0.602 ± 0.027

0.751 ± 0.028

0.773 ± 0.016

 

0.764 ± 0.017

 

0.775 ± 0.019

CNN-LSTM

0.663 ± 0.028

0.780 ± 0.003

0.770 ± 0.009

 

0.772 ± 0.004

 

0.779 ± 0.018

Back to article page

BMC Medical Informatics and Decision Making

ISSN: 1472-6947

Contact us