Table 1.
Methods performance metrics [accuracy and 95% confidence interval (CI), sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), kappa, F1 score and Matthew's Correlation Coefficient (MCC)] on external validation set. Prediction models were developed within different methods: Linear Discriminant Analysis (LDA), Generalized Linear Model (GLM) with logit link function, Naïve Bayes (NB), Classification and Regression Trees (CART), k-Nearest Neighbors (kNN), Support Vector Machines (SVM), Random Forest (RF) and Neural Network (NN).
| Method | Accuracy | 95% CI | Se | Sp | PPV | NPV | kappa | F1 score | MCC |
|---|---|---|---|---|---|---|---|---|---|
| LDA | 0.797 | 0.784–0.810 | 0.524 | 0.909 | 0.703 | 0.823 | 0.468 | 0.600 | 0.478 |
| GLM | 0.796 | 0.783–0.809 | 0.511 | 0.913 | 0.706 | 0.820 | 0.461 | 0.593 | 0.472 |
| NB | 0.753 | 0.739–0.767 | 0.593 | 0.819 | 0.573 | 0.831 | 0.407 | 0.583 | 0.408 |
| CART | 0.793 | 0.780–0.806 | 0.616 | 0.865 | 0.652 | 0.846 | 0.490 | 0.634 | 0.490 |
| kNN | 0.754 | 0.740–0.768 | 0.381 | 0.907 | 0.626 | 0.781 | 0.325 | 0.474 | 0.342 |
| SVM | 0.791 | 0.777–0.804 | 0.479 | 0.919 | 0.706 | 0.812 | 0.439 | 0.571 | 0.454 |
| RF | 0.797 | 0.783–0.810 | 0.553 | 0.897 | 0.687 | 0.830 | 0.477 | 0.613 | 0.482 |
| NN | 0.796 | 0.782–0.808 | 0.567 | 0.889 | 0.677 | 0.834 | 0.479 | 0.617 | 0.482 |