Table 6.

Top 5 Predictive variables for all models predicting mortality, LOS, and discharge disposition.

Top five predictive variables
Died
Neural Network	Total Number of Diagnoses
	APRDRG- Mortality
	APRDRG- Severity
	Age
	Census Division of Hospital
Random Forrest	Age
	Total Number of Diagnoses
	Month of Procedure
	Hospital Location/teaching status
	Patient Location: NCHS Urban-Rural Code
XGBoost	Age
	Total Number of Diagnoses
	Race
	APRDRG- Severity
	Median household income national quartile for patient ZIP Code
LSVM	Total Number of Diagnose
	APRDRG- Mortality
	Age
	Census Division of Hospital
	Month of Procedure
Discharge
Neural Network	Age
	Total Number of Diagnoses
	APRDRG- Severity
	Region of Hospital
	Primary expected payer
Random Forrest	Age
	Total Number of Diagnoses
	Hospital Bedsize
	Median household income national quartile for patient ZIP Code
	Sex
XGBoost	Age
	Total Number of Diagnoses
	Sex
	Race
	Region of Hospital
LSVM	Age
	Total Number of Diagnoses
	Region of Hospital
	Sex
	Race
LOS
Neural Network	Total Number of Diagnoses
	Age
	APRDRG- Mortality
	APRDRG- Severity
	Primary expected payer
Random Forrest	Age
	Total Number of Diagnoses
	Median household income national quartile for patient ZIP Code
	Hospital Bedsize
	Sex
XGBoost	Total Number of Diagnoses
	Age
	Hospital Control
	Region of Hospital
	Patient Location: NCHS Urban-Rural Code
LSVM	Total Number of Diagnoses
	Age
	APRDRG- Mortality
	APRDRG- Severity
	Census Division of Hospital