Table 3 Comparison of sensitivity of GLAM and GMIC models with (columns indicated as “Transfer”) and without (columns indicated as “Pre-trained”) transfer learning in four concordance levels of Dataset 2.
Concordance LEVEL | GLAM, Pre-TRAINED | GMIC, Pre-TRAINED | GLAM, TRANSFER | GMIC, TRANSFER |
|---|---|---|---|---|
Almost Perfect | 82.40% | 83.70% | 93.20% | 95.50% |
Substantial | 81.00% | 83.10% | 92.00% | 94.60% |
Moderate | 79.20% | 80.50% | 90.90% | 93.40% |
Poor | 76.90% | 76.80% | 87.50% | 90.80% |
p-value | 0.029 | 0.014 | 0.021 | 0.038 |
