Table 6.
Comparison of the average and weighted performance metrics using logistic regression-based nomogram for the external datasets.
| Overall |
Weighted with 95% CI |
|||||
|---|---|---|---|---|---|---|
| Dataset | Country | Accuracy | Precision | Recall | F1-score | Specificity |
| Internal validation | Italy (OSR) | 91 ± 1.41 | 89.5 ± 1.49 | 91.12 ± 1.38 | 90.3 ± 1.44 | 89.71 ± 1.54 |
| External validation | Italy-1 | 91.69 ± 1.93 | 88.02 ± 2.27 | 97.13 ± 1.17 | 92.35 ± 1.86 | 85.89 ± 2.43 |
| Italy-2 | 95.18 ± 1.98 | 94.63 ± 2.08 | 97.24 ± 1.51 | 95.92 ± 1.83 | 92.31 ± 2.46 | |
| Italy-3 | 92.86 ± 1.97 | 88.79 ± 2.42 | 97.17 ± 1.27 | 92.79 ± 1.98 | 88.98 ± 2.4 | |
| Brazil-1 | 92.7 ± 2.75 | 94.25 ± 2.46 | 95.26 ± 2.24 | 93.25 ± 2.65 | 87.27 ± 3.52 | |
| Brazil-2 | 95.11 ± 1.22 | 94.42 ± 2.06 | 94.25 ± 2.12 | 94.33 ± 2.14 | 95.67 ± 2.11 | |
| Ethiopiaa | 92.5 ± 3.65 | – | 92.5 ± 3.65 | – | – | |
| Average | 93.34 ± 2.19 | 92.02 ± 2.26 | 95.59 ± 2.18 | 93.73 ± 2.33 | 90.54 ± 2.47 | |
a
Ethiopia dataset has only COVID patients.