Skip to main content
. 2021 Aug 2;11:15619. doi: 10.1038/s41598-021-95114-3

Table 2.

Features a priori considered and features selected by GLMNET for their relevance in predicting ICU admission in five models.

Model Features Selected features (ICU admission)
Blood laboratory and arterial gas analyses Age, LDH, D-dimer, PCR, Lymphocytes, P/F Age, LDH, P/F
Radiological %consolid, %ground glass, %normal lung %consolid, %normal lung
Radiomics 86 radiomic features (see Supplementary Table 1) Dependence Non Uniformity, Small Dependence High Gray Level Emphasis, Large Dependence Low Gray Level Emphasis, Correlation, Interquartile Range, Total Energy, Run Variance, Large Area Low Gray Level Emphasis, Small Area Low Gray Level Emphasis
Hybrid radiological Age, LDH, D-dimer, PCR, Lymphocytes, P/F, %consolid, %ground glass, %normal lung Age, P/F, LDH, %consolid
Hybrid radiomics Age, LDH, D-dimer, PCR, Lymphocytes, P/F, %consolid, %ground glass, %normal lung, 86 radiomic features P/F, LDH, %consolid, Large Dependence Low Gray Level Emphasis, Run Length Non Uniformity, Low Gray Level Zone Emphasis

P/F = ratio between the arterial partial pressure of oxygen [PaO2] measured (in mmHg) by blood gas analysis and fraction of inspired oxygen [FiO2]; %consolid = percentage of consolidated lung; %ground glass = percentage of lung exhibiting ground glass opacities density; %normal lung = percentage of lung with normal density.

CRP serum C-reactive protein, ICU intensive care unit, LDH serum lactate dehydrogenase.