Table 5.
Confusion matrix for predicting the SVM classifier for the test set (n = 383)
| Prediction | True entity | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Esophagus | Larynx | Lung | Lymphoma | Oral cavity | Oropharynx | Skin | Balanced Accuracy | Sensitivity | Specificity | |
| Esophagus | 24 | 1 | 1 | 0 | 1 | 1 | 0 | 0.87 | 0.75 | 0.99 |
| Larynx | 3 | 31 | 2 | 0 | 5 | 2 | 1 | 0.86 | 0.76 | 0.96 |
| Lung | 1 | 3 | 132 | 0 | 3 | 0 | 0 | 0.97 | 0.98 | 0.97 |
| Lymphoma | 0 | 0 | 0 | 16 | 0 | 0 | 0 | 1.00 | 1.00 | 1.00 |
| Oral cavity | 3 | 5 | 0 | 0 | 97 | 5 | 0 | 0.90 | 0.85 | 0.95 |
| Oropharynx | 1 | 1 | 0 | 0 | 7 | 25 | 1 | 0.86 | 0.76 | 0.97 |
| Skin | 0 | 0 | 0 | 0 | 1 | 0 | 10 | 0.92 | 0.83 | 1.00 |
The true entity of the cohorts’ samples is displayed, and correct predictions are marked in bold. The balanced accuracy is shown as a performance parameter in single-class prediction