Skip to main content
. 2021 May 29;134:104531. doi: 10.1016/j.compbiomed.2021.104531

Table 4.

Important biomarkers in machine learning models for the diagnosis and classification of COVID-19 severity.

Biomarkers Diagnostic model
 Disease severity model
ANN DT PLSDA KNN ANN DT PLSDA KNN
Ferritin + + + + + + + + + + + + + + + +
Gamma-glutamyltransferase + - - - - - - -
HCO3 (arterial) + - - + - - + +
Base excess (arterial) + - - - - - + +
Base excess (venous) - - - - - - + -
Sodium + - - - - - - -
Total O2 (arterial) - - - - + - - -
pO2 (arterial) - - - + - - - -
Total CO2 - + + + - - - +
pCO2 (arterial) + - - - - - + +
pCO2 (venous) + - + - - - + -
Indirect bilirubin + - - - - - - -
Alkaline phosphatase + - - - + - - -
Urine pH - - + + + - + -
pH (venous) - - - - - - + -
pH (arterial) - - - - - - - +
FiO2 (arterial) - - - - + - -
ctO2 (arterial) - - - - - - - +
Total bilirubin - - - - + - - -
Red blood cell distribution width - - - - + - - -
Platelets - - - - + - - -
C-reactive protein - - - - + - + -
Calcium ionised - - + + - - - -
Urine-density - - + + - - - -
Lactate dehydrogenase - - - - - - + -
Arterial lactic acid - - - + - - - +
Haemoglobin saturation (arterial) - - - - - - - +
Phosphorous - - + + - - - -
Lipase dosage - - - - - - + -
Rods - - - + - - + -

Less important variable (−); Important variable (+); Critical variable (++).