Skip to main content

Table 2 Predictor variables used in this study

From: Using machine learning methods to predict in-hospital mortality of sepsis patients in the ICU

Predictors
Acute physiology (first 24 h in the ICU) Chronic health status
 Heart rate*  Elixhauser comorbidity index
 Systolic blood pressure*  Congestive heart failure
 Diastolic blood pressure*  Cardiac arrhythmias
 Mean blood pressure*  Valvular heart disease
 Respiratory rate*  Pulmonary circulation
 Temperature*  Peripheral vascular
 SpO2* (blood oxygen saturation)  Hypertension
 Total CO2*  Other neurological diseases
 pCO2* (partial pressure of CO2)  Chronic obstructive pulmonary disease
 pH* (acidity in the blood)  Diabetes without complications
 Urine output  Diabetes with complications
 Glasgow Coma Score (GCS)  Hypothyroidism
 GCS (eye)  Renal failure
 GCS (motor)  Liver disease
 GCS (verbal)  Metastatic cancer
 Anion gap*  Coagulopathy
 Bicarbonate*  Obesity
 Creatinine*  Fluid electrolyte
 Chloride*  Alcohol abuse
 Glucose*  Depression
 Haematocrit*  Renal replacement therapy
 Haemoglobin* Other
 Lactate*  Gender
 Platelet*  Weight loss
 Potassium*  Ventilation
 Partial thromboplastin time*  Age
 INR*  Weight
 Prothrombin time*  SAPS II score (first 24 h in the ICU)
 Sodium*  SOFA score (first 24 h in the ICU)
 Blood urea nitrogen (BUN)*  
 WBC*  
 Acute kidney injury  
  1. *: each predictor marked with * means that it is a time-stamped variable, and its corresponding minimum and maximum values within the first 24 h in the ICU were used as inputs in model development