Table 2 Pooled analysis of other evaluation metrics (accuracy, sensitivity, specificity) for mortality prediction
Model | Accuracy | Sensitivity | Specificity | ||||||
|---|---|---|---|---|---|---|---|---|---|
N | Pooled AUC | Confidence interval | N | Pooled AUC | Confidence interval | N | Pooled AUC | Confidence interval | |
Random forest | 15 | 0.600 | 0.415–0.786 | 12 | 0.619 | 0.547–0.691 | 8 | 0.755 | 0.704–0.806 |
Logistic regression | 10 | 0.576 | 0.366–0.786 | 9 | 0.592 | 0.508–0.676 | 7 | 0.745 | 0.685–0.805 |
Support vector machine | 8 | 0.743 | 0.694–0.791 | 7 | 0.590 | 0.492–0.687 | 5 | 0.761 | 0.572–0.950 |
Gradient boosting | 12 | 0.706 | 0.467–0.945 | 12 | 0.678 | 0.527–0.829 | 8 | 0.757 | 0.664–0.850 |
Neural Network | 4 | 0.441 | 0.061–0.820 | 3 | 0.623 | 0.503–0.743 | 3 | 0.774 | 0.657–0.891 |
Decision tree | 2 | 0.810 | 0.504–1.117 | 3 | 0.609 | 0.224–0.994 | 1 | NA | NA |
K-nearest neighbors | 2 | 0.601 | 0.576–0.627 | 2 | 0.339 | 0.314–0.364 | 2 | 0.685 | 0.592–0.778 |
Lasso regression | 4 | 0.413 | 0.079–0.747 | 2 | 0.776 | 0.564–0.988 | 2 | 0.701 | 0.610–0.793 |
Bayesian network | 1 | NA | NA | 0 | NA | NA | 0 | NA | NA |