Predicting Long-Term Mortality after Acute Coronary Syndrome Using Machine Learning Techniques and Hematological Markers

Table 1

Variables used in the Cox regression model, machine-learned model, and for the calculation of the GRACE score.

COX regression	Machine-learned model	GRACE score
(310 observations excluded due to missing values)	(84 observations excluded due to missing values that where required to calculate GRACE score)
Troponin elevation ratio	Troponin elevation ratio	Age
Neutrophil to lymphocyte ratio; Platelet to lymphocyte ratio	Red cell distribution width	Heart rate
Red cell distribution width	Platelet count	Systolic blood pressure
Platelet count	Creatinine level	Creatinine level
Creatinine level	Hemoglobin level	ST-segment deviation
Fibrinogen level	Mean cell volume	Troponin elevation (true or false)
Hemoglobin level	Sodium level	Killip class
Potassium level	Prothrombin time
Mean cell volume	Fibrinogen level
Monocyte count	Age
Sodium level	Lymphocyte count
Prothrombin level	Neutrophil count
Age	LDL level
Heart rate at admission	CRP level
Systolic blood pressure	Sex
ST-segment elevation	Heart rate
Diabetes	Systolic blood pressure
	Diastolic blood pressure
	Body mass index

LDL: low-density lipoprotein; CRP: C-reactive protein.