Table 4.
Statistical comparison of the test set’s cross-modal loss with the best DACMVA model compared to TDImpute. The best DACMVA model is shown in Table 3.
| Missing % | Best DACMVA Models vs. TDImpute |
|---|---|
| 4% | p-value = 1.2 × 10−10 |
| 90% | p-value = 6.8 × 10−27 |
| 95% | p-value = 3.3 × 10−44 |