Table 3.
The results of subgroup analysis.
| Subgroup | Number of study | Sensitivity (95 % CI) |
I2 (%) |
Specificity | I2 (%) |
PLR | I2 (%) |
NLR | I2 (%) |
AUC |
|---|---|---|---|---|---|---|---|---|---|---|
| Imaging modality | ||||||||||
| CRX | 4 | 0.91(0.88,0.94) | 85.6 | 0.96(0.95,0.98) | 95.3 | 26.04(3.73,181.94) | 93.3 | 0.04(0.00,0.41) | 92.6 | 0.9914 |
| CT | 28 | 0.89(0.88,0.90) | 78.9 | 0.89(0.87,0.90) | 62.1 | 6.92(5.35,8.96) | 69.5 | 0.14(0.11,0.19) | 80.0 | 0.9427 |
| Modeling methods | ||||||||||
| Radiomic algorithm | 13 | 0.92(0.90,0.94) | 78.4 | 0.90(0.87,0.92) | 36.8 | 7.16(4.96,10.33) | 53.0 | 0.15(0.08,0.28) | 85.6 | 0.9446 |
| Deep learning | 19 | 0.88(0.87,0.89) | 78.0 | 0.91(0.90,0.92) | 88.5 | 8.32(5.69,12.18) | 82.5 | 0.12(0.09,0.17) | 76.9 | 0.9702 |
| sample size | ||||||||||
| ＜100 | 18 | 0.87(0.83,0.90) | 65.4 | 0.89(0.86,0.92) | 47.8 | 6.50(4.42,9.58) | 49.3 | 0.18(0.12,0.28) | 59.0 | 0.9371 |
| ＞100 | 14 | 0.89(0.88,0.90) | 87.0 | 0.91(0.90,0.92) | 90.8 | 8.81(6.02,12.89) | 86.2 | 0.10(0.07,0.14) | 88.6 | 0.9725 |
| ROI | ||||||||||
| Infection regions | 15 | 0.89(0.88,0.90) | 81.0 | 0.89(0.88,0.91) | 48.8 | 6.89(5.20,9.12) | 58.0 | 0.14(0.09,0.20) | 81.3 | 0.9409 |
| others | 16 | 0.88(0.86,0.90) | 80.4 | 0.92(0.90,0.94) | 89.5 | 9.33(5.64,15.45) | 83.3 | 0.11(0.07,0.19) | 83.2 | 0.9691 |
| segmentation | ||||||||||
| 2D | 14 | 0.91(0.89,0.93) | 71.6 | 0.93(0.91,0.95) | 88.9 | 9.71(5.78,16.33) | 79.3 | 0.10(0.06,0.17) | 77.3 | 0.9740 |
| 3D | 15 | 0.88(0.87,0.90) | 85.1 | 0.89(0.87,0.90) | 64.8 | 6.77(4.79,9.57) | 76.6 | 0.15(0.10,0.22) | 85.9 | 0.9386 |
Abbreviations: AUC: area under the curve; CT: Computed tomography; CXR: Chest X-Ray; NLR: negative likelihood ratio; PLR:positive likelihood ratio; ROI: Region of interest;2D: two-dimensional;3D: three-dimensional