Table 4. Performance of the long-term outcome prediction models.

	Calibration (HL test)		Discrimination
Models	C & S R2	N R2	Sen	Spec	Youden	AUROC
Predict Unfavorable/Favorable
Age, GCS ≦ 8	0.32	0.45	0.77	0.76	0.53	0.81
Age, GCS ≦ 8, SDNN, RMSSD, Ln(TP), Ln(VLF), Ln(LF), Ln(HF)	0.42	0.58	0.89	0.78	0.67	0.91
Predict Dead/Alive
Age, GCS ≦ 8	0.20	0.29	0.55	0.85	0.40	0.74
Age, GCS ≦ 8, SDNN, RMSSD, Ln(TP), Ln(VLF), Ln(LF), Ln(HF)	0.33	0.48	0.76	0.82	0.56	0.86

HL test: Hosmer-Lemeshow goodness of fit test; C & S R²: Cox and Snell R²; N R²: Nagelkerke R²; Sen: Sensitivity; Spec: specificity; AUROC: area under receiver operating characteristic curve