Table 4 (Ternary) CNN accuracy, sensitivity and specificity at the patient level of the ternary (COV+ vs COV− vs CTRL) classification model using leave-one patient out cross-validation, with the related COV+ sensitivity and specificity percentages. (Binary) CNNs accuracy, sensitivity and specificity at the patient level for the binary (COV+ vs CTRL, COV+ vs COV− and COV− vs CTRL) classification model.
Accuracy (%) | Sensitivity (%) | Specificity (%) | COV+ sensitivity (%) | COV+ specificity | |
|---|---|---|---|---|---|
Ternary | |||||
Patient-level | 89 | 89 | 95 | 83.3 | 96 |
Patient-level (COV+ = 26) | 92 | 91 | 96 | 85 | 96 |
Binary | |||||
COV+ vs CTRL | 89 | 83 | 94 | – | – |
COV+ vs COV− | 89.6 | 83 | 93 | – | – |
COV− vs CTRL | 91.5 | 89 | 94 | – | – |
