Table 3.
The performance of radiomics auto-ML model in COVID-19 pneumonia 1st CT images.
| TPR (%) | TNR (%) | PPV (%) | NPV (%) | AUC | P-value | |
|---|---|---|---|---|---|---|
| Moderate and severe | ||||||
| Training(n = 175) | 98.1(107/109) | 98.4(65/66) | 99.0(107/108) | 97.0(65/67) | 0.98 | 0.0001 |
| Test(n = 44) | 96.2(26/27) | 94.1(16/17) | 96.2(26/27) | 94.1(16/17) | 0.95 | 0.000 |
| Moderate and control | ||||||
| Training(n = 189) | 96.2(76/79) | 98.1(107/109) | 97.4(76/78) | 97.2(107/110) | 0.97 | 0.000 |
| Test(n = 47) | 100(20/20) | 96.2(26/27) | 95.2(20/21) | 100(26/26) | 0.98 | 0.000 |
| Severe and control | ||||||
| Training(n = 145) | 98.7(79/80) | 98.4(64/65) | 98.7(79/80) | 98.4(64/65) | 0.99 | 0.000 |
| Test(n = 38) | 94.7(18/19) | 94.4(17/18) | 94.7(18/19) | 94.4(17/18) | 0.95 | 0.0001 |