Table 4.

Prediction results using receiver operating characteristic (ROC) analysis

	Machine learning 1,2	Logistic regression 3,4
AUC [95% CI]	0.98 [0.95-1]	0.93 [0.87-0.99]
Cutoff score	49	0.25
Sensitivity [95% CI]	94.9% [87.9%-100%]	82.1% [70.0%-94.1%]
Specificity [95% CI]	97.4% [95.0%-99.9%]	92.3% [88.1%-96.5%]
PPV [95% CI]	90.2% [81.2%-99.3%]	72.7% [59.6%-85.9%]
NPV [95% CI]	98.7% [96.9%-100%]	95.4% [92.0%-98.7%]

AUC: area under the curve; CI: confidence interval; PPV: positive predictive value; NPV: negative predictive value.

¹The range of machine learning score is [0, 100].

²Variables used in the machine learning method were road traffic accident, history of loss of consciousness, vomiting, seizure activity, confusion, clinical signs of skull fracture, and signs of base of skull fracture.

³The range of logistic regression score is [0, 1].

⁴Variables used in the logistic regression model were road traffic accident, history of loss of consciousness, vomiting, and signs of base of skull fracture.