. Author manuscript; available in PMC: 2018 Apr 4.

Published in final edited form as: Acad Emerg Med. 2016 Feb 13;23(3):269–278. doi: 10.1111/acem.12876

Table 3.

AUC Comparison of Random Forest to Other Models

Model	Observations^*	AUC	SE	Lower 95% CI	Upper 95% CI	Prob > χ²^†
Random forest	1,054	0.860	0.021	0.819	0.900	N/A
Logistic regression	1,054	0.755	0.034	0.689	0.821	0.003
MEDS score	1,052	0.705	0.032	0.634	0.765	<0.0001
CURB-65 score	1,054	0.734	0.032	0.670	0.797	<0.0001
REMS score	1,052	0.717	0.035	0.649	0.785	<0.0001
CART model	1,054	0.693	0.037	0.62	0.766	<0.0001

AUC = area under the curve; CART = classification and regression tree; CURB-65 = Confusion, Urea nitrogen, Respiratory rate, Blood pressure, ≥65 years; MEDS = Mortality in ED Sepsis; REMS = Rapid Emergency Medicine Score.

Unable to compute MEDS and REMS score in two patients secondary to missing data

^†

Compared to Random forest model.