Table 3. Performance of the prediction model in the training and testing sets.

Datasets	AUC	Accuracy	Sensitivity	Specificity	PPV	NPV
Our model
Training set	0.829 (0.768–0.889)	0.773 (0.710–0.827)	0.783 (0.719–0.846)	0.740 (0.618–0.862)	0.514 (0.398–0.629)	0.906 (0.858–0.955)
Testing set	0.825 (0.728–0.923)	0.747 (0.645–0.833)	0.677 (0.563–0.791)	0.923 (0.821–1.000)	0.533 (0.388–0.679)	0.957 (0.898–1.000)
SOFA
Training set	0.732 (0.649–0.814)	0.692 (0.625–0.754)	0.677 (0.605–0.749)	0.740 (0.618–0.862)	0.416 (0.313–0.518)	0.893 (0.839–0.948)
Testing set	0.707 (0.584–0.830)	0.747 (0.645–0.833)	0.862 (0.778–0.946)	0.462 (0.270–0.653)	0.571 (0.360–0.783)	0.800 (0.706–0.894)
APACHE II
Training set	0.713 (0.634–0.791)	0.678 (0.610–0.740)	0.665 (0.592–0.738)	0.720 (0.596–0.844)	0.400 (0.299–0.501)	0.884 (0.827–0.941)
Testing set	0.718 (0.601–0.835)	0.670 (0.564–0.765)	0.631 (0.513–0.748)	0.769 (0.607–0.931)	0.455 (0.307–0.602)	0.872 (0.777–0.968)

Notes.

SOFA

Sepsis Organ Failure Assessment

APACHE II

Acute Physiology and Chronic Health Evaluation II

AUC

area under the receiver operating characteristic curve

PPV

positive predictive value

NPV

negative predictive value

Our model: integrated systolic blood pressure, cerebrovascular disease, Na, oxygenation index (PaO2/FiO2), prothrombin time, glucocorticoids, and hemodialysis;

SOFA: a quantitative scoring index that dynamically characterizes sepsis-associated organ dysfunction, encompassing respiratory system, coagulation system, liver function, cardiovascular system, central nervous system, and renal function.

APACHE II: incorporated the age score, acute physiology score, and chronic health status score.