TABLE 3.
Comparing the performance error prediction using spectral entropy features vs. number of autocorrelation peaks.
| Cross validation (3-fold) performance metric | H5% |
| Specificity | 0.79 |
| Precision | 0.88 |
| Recall | 0.79 |
| F1 | 0.84 |
| ROC-AUC | 0.76 |
| HPER (Etotal ≥ 3%) | 14 |
| LPER (Etotal < 3%) | 29 |
Measures in bold have significant p values.