Table 5.
Classification and regression results of multi-task transfer learning using CTP text. AD: Alzheimer’s Disease. Accuracy: mean and standard deviation of results of 5 rounds. Best: highest accuracy of all epochs in 5 rounds. RMSE: mean and standard deviation of Root-Mean-Square Errors of 5 rounds. Best RMSE: lowest RMSE of all epochs in 5 rounds.
| Model | Pre-training | Settings | Accuracy % | Best % | RMSE | Best RMSE |
|---|---|---|---|---|---|---|
| Text (Luz et al, 2020) | - | Classification | 75 | - | ||
| - | Regression | 5.20 | - | |||
| BERT base | No | Classification | 76.67 ± 1.56 | 81.25 | - | - |
| Regression | - | - | 5.18 ± 0.04 | 4.65 | ||
| Multi-task | 78.75 ± 1.56 | 83.33 | 4.70 ± 0.02 | 4.39 | ||
| Yes | Classification | 80.83 ± 2.04 | 85.42 | - | - | |
| Regression | - | - | 4.15 ± 0.01 | 4.06 | ||
| Multi-task | 80.83 ± 1.56 | 87.50 | 4.96 ± 0.01 | 4.20 |