Table 3.
AUC values of COVID-19 prognosis experiments, which aim to predict the survival of COVID-19 patients.
| Methods | Year | Ratio of training data | Retrospective studies | Prospective studies | ||||
|---|---|---|---|---|---|---|---|---|
| Image-only | Text-only | Image-text | Image-only | Text-only | Image-text | |||
| CLIP54 | 2021 | 1% | 70.4 | 84.3 | 89.5 | 66.9 | 76.7 | 81.3 |
| ConVIRT34 | 2022 | 1% | 75.3 | 88.1 | 92.6 | 70.6 | 81.2 | 85.4 |
| BioViL30 | 2022 | 1% | 77.1 | 89.0 | 92.9 | 70.8 | 82.1 | 85.7 |
| Med-MLLM | Ours | 1% | 82.8(0.5) | 92.1(0.3) | 95.7(0.3) | 81.2(0.7) | 88.3(0.8) | 92.0(0.5) |
| CLIP54 | 2021 | 100% | 79.5 | 91.7 | 93.2 | 70.6 | 84.0 | 88.3 |
| ConVIRT34 | 2022 | 100% | 83.4 | 93.8 | 95.4 | 77.5 | 88.7 | 90.1 |
| BioViL30 | 2022 | 100% | 83.5 | 94.2 | 95.1 | 77.0 | 87.9 | 89.8 |
| Med-MLLM | Ours | 100% | 88.1(0.2) | 95.3(0.2) | 96.6(0.1) | 85.7(0.4) | 92.8(0.2) | 94.9(0.2) |
All values are reported in percentage (%). The best results are in bold.