Table 1.

Model performance in derivation and validation datasets for, partial least squares (PLS), support vector machines linear and kernel (SVM-L and SVM-K).

Features	Derivation dataset (median AUC of 100 runs)			Validation dataset [AUC (95% confidence intervals)]
	Classifiers			Classifiers
	PLS	SVM-L	SVM-K	PLS	SVM-L	SVM-K
Age	0.58	0.54	0.53	0.58 (0.46–0.7)	0.6 (0.48–0.71)	0.64 (0.53–0.76)
Sex	0.59	0.59	0.59	0.62 (0.5–0.74)	0.62 (0.5–0.74)	0.62 (0.5–0.74)
Hunt Hess Scale	0.60	0.55	0.58	0.49 (0.37–0.61)	0.46 (0.34–0.58)	0.5 (0.38–0.62)
Modified Fisher Scale	0.54	0.57	0.50	0.47 (0.35–0.59)	0.53 (0.41–0.65)	0.53 (0.41–0.65)
Glasgow Coma Scale	0.59	0.57	0.63	0.43 (0.31–0.55)	0.44 (0.32–0.56)	0.56 (0.44–0.68)
Baseline (age, sex, and scales)	0.63	0.58	0.54	0.64 (0.53–0.76)	0.59 (0.48–0.71)	0.61 (0.49–0.72)
Diastolic blood pressure	0.52	0.48	0.56	0.44 (0.26–0.61)	0.42 (0.25–0.59)	0.56 (0.19–0.53)
Systolic blood pressure	0.58	0.54	0.49	0.65 (0.49–0.82)	0.43 (0.26–0.6)	0.36 (0.19–0.53)
Heart rate	0.55	0.51	0.50	0.46 (0.28–0.63)	0.5 (0.33–0.68)	0.45 (0.28–0.62)
Oxygen saturation	0.56	0.53	0.50	0.62 (0.45–0.79)	0.48 (0.31–0.65)	0.5 (0.33–0.67)
Respiratory rate	0.49	0.50	0.50	0.57 (0.4–0.74)	0.54 (0.36–0.71)	0.5 (0.33–0.67)
Combined physiological	0.66	0.56	0.50	0.47 (0.3–0.64)	0.51 (0.34–0.68)	0.5 (0.33–0.67)
Baseline and physiological	0.63	0.56	0.50	0.5 (0.33–0.67)	0.5 (0.33–0.67)	0.5 (0.33–0.67)
MRMR (baseline and physiological)	0.71	0.60	0.50	0.78 (0.64–0.92)	0.64 (0.47–0.8)	0.5 (0.33–0.67)

The SVM-L classifier with maximal relevance and minimal redundancy (MRMR) feature reduction performed the best.

Values highlighted in bold indicates the performance of the classifier that performed the best.