Table 2.
Model accuracy for Patients 2 (word reading) and 5 (picture naming).
| Accuracy (%) | STFM+/ESM+ | STFM−/ESM+ | STFM−/ESM− | STFM+/ESM− | Sensitivity (%) | Specificity (%) | AUC | p-value | Equation | |
|---|---|---|---|---|---|---|---|---|---|---|
| Patient 2 | ||||||||||
| HG all duration | 76.5% | 8 | 3 | 5 | 1 | 72.7% | 83.3% | 0.79 | 0.0465 | (1) |
| HG PCA selected duartion | 76.5% | 8 | 3 | 5 | 1 | 72.7% | 83.3% | 0.7 | 0.0439 | (2) |
| HG scaled by centrality | 82.4% | 10 | 1 | 4 | 2 | 90.9% | 66.7% | 0.83 | 0.134 | (3) γ3 = 0 |
| HG + CCEPs z-score | 82.4% | 9 | 2 | 5 | 1 | 81.8% | 83.3% | 0.83 | 0.103 | (3) CCEPs = z-score |
| HG + CCEPs edges | 82.4% | 9 | 2 | 5 | 1 | 81.8% | 83.3% | 0.86 | 0.0747 | (3) CCEPs = Edges |
| Patient 5 | ||||||||||
| HG all duration | 52.9% | 9 | 1 | 0 | 7 | 90.0% | 0.0% | 0.31 | 0.881 | (1) |
| HG PCA selected duartion | 58.8% | 8 | 2 | 2 | 5 | 80.0% | 28.6% | 0.39 | 0.952 | (2) |
| HG scaled by centrality | 82.4% | 8 | 2 | 6 | 1 | 80.0% | 85.7% | 0.73 | 0.947 | (3) γ3 = 0 |
| HG scaled by CCEPs z-score | 82.4% | 8 | 2 | 6 | 1 | 80.0% | 85.7% | 0.8 | 0.0673 | (3) CCEPs = z-score |
| HG scaled by CCEPs edges | 88.2% | 9 | 1 | 6 | 1 | 90.0% | 85.7% | 0.81 | 0.0462 | (3) CCEPs = Edges |
Accuracy (%), areas under ROC curve (AUC), and p-values were calculated using Classification Learner from Matlab. Sensitivity was calculated using STFM+ESM+ and STFM−ESM+ electrode numbers, and specificity using STFM−ESM− and STFM+ESM− electrodes. Results for Equation 1 are listed on the first row of each subtable; results for Equation 2 are listed on the second row; results for Equation 3 are listed in several steps on the rest of the rows.