Fig. 4
From: Intelligent prediction of RBC demand in trauma patients using decision tree methods
Feature importance ranked by weight in the transfusion prediction model. a Non-invasive parameters for prediction. b All parameters for prediction. Trauma_loc trauma location, SBP systolic blood pressure, SI shock index, DBP diastolic blood pressure, HR, heart rate, R respiration, T temperature, SpO2 blood oxygen saturation, Hct hematocrit, TCO2 total carbon dioxide pH potential of hydrogen, PCO2 partial pressure of carbon dioxide, CRP C-reactive protein, VD vasoactive drugs, Fib fibrinogen, SB standard bicarbonate, SCFL severity classification in first level, Lac lactate, PO2 partial pressure of oxygen, AB actual bicarbonate, PLT platelet count, EDT emergency department time, INR international standardized ratio