Table 2 Model performances of the best model based on electroencephalography features, mean arterial pressure and age, the best model including IMPACT features at admission and the IMPACT predictor independently, predicting poor outcome both as Extended Glasgow Outcome Scale (GOSE) 1–2 and GOSE 1–4. For both the internal validation within the training set and the validation set, the number of patients (N), area under the receiver operating characteristic curve (AUC) values with 50% confidence interval, and sensitivity and specificity at an optimal threshold are shown
Performance measures for the prediction model | Best model, 72 + 96 h | Best model, 72 + 96 h + IMPACT | IMPACT predictor | |||
|---|---|---|---|---|---|---|
Poor outcome | GOSE 1–2 | GOSE 1–4 | GOSE 1–2 | GOSE 1–4 | GOSE 1–2 | GOSE 1–4 |
Training set | ||||||
 N | 30 | 25 | 29 | 25 | 38 | 30 |
 AUC (50% CI) | 0.88 (0.83–0.91) | 0.66 (0.57–0.74) | 0.94 (0.89–0.96) | 0.61 (0.51–0.68) | 0.74 (0.66–0.79) | 0.67 (0.61–0.73) |
 Threshold | 0.541 | 0.740 | 0.680 | 0.591 | 0.650 | 0.530 |
 Sensitivity | 0.92 (0.81–0.98) | 0.71 (0.59–0.80) | 0.75 (0.61–0.85) | 0.76 (0.66–0.85) | 0.65 (0.54–0.75) | 0.63 (0.53–0.72) |
 Specificity | 0.77 (0.66–0.85) | 0.75 (0.57–0.88) | 1.00 (0.92–1.00) | 0.50 (0.33–0.67) | 0.81 (0.72–0.88) | 0.82 (0.67–0.91) |
Validation set | ||||||
 N | 14 | 14 | 14 | 14 | 19 | 19 |
 AUC (50% CI) | 0.75 (0.60–0.86) | 0.73 (0.62–0.82) | 0.81 (0.69–0.88) | 0.76 (0.65–0.89) | 0.84 (0.76–0.89) | 0.87 (0.78–0.92) |
 Threshold | 0.623 | 0.805 | 0.378 | 0.668 | 0.550 | 0.349 |
 Sensitivity | 0.83 (0.61–0.95) | 0.78 (0.61–0.89) | 1.00 (0.79–1.00) | 0.89 (0.73–0.97) | 0.88 (0.70–0.96) | 1.00 (0.88–1.00) |
 Specificity | 0.88 (0.70–0.96) | 0.80 (0.55–0.94) | 0.75 (0.57–0.88) | 0.80 (0.55–0.94) | 0.73 (0.58–0.84) | 0.75 (0.57–0.88) |