Table 5.
Performance comparison between different models. P-values are calculated using a paired t-test between the proposed and each competing method.
| Model | accuracy (p-value) | sensitivity (p-value) | specificity (p-value) |
|---|---|---|---|
| (a) 6 month prediction. | |||
| Proposed | 0.81 | 0.84 | 0.80 |
| Baseline | 0.76 (1.16e-08) | 0.81 (0.07) | 0.75 (1.87e-10) |
| Single modal (cognitive performance) | 0.74 (3.17e-09) | 0.81 (8.12e-05) | 0.70 (2.68e-14) |
| Single modal (CSF) | 0.72 (4.65e-17) | 0.78 (8.12e-05) | 0.70 (2.68e-14) |
| Single modal (MRI) | 0.70 (4.68e-24) | 0.75 (6.71e-10) | 0.68 (2.05e-21) |
| (b) 12 month prediction | |||
| Proposed | 0.81 | 0.84 | 0.80 |
| Baseline | 0.76 (1.77e-07) | 0.82 (0.057) | 0.75 (3.65e-07) |
| Single modal (cognitive performance) | 0.75 (2.70e-08) | 0.78 (0.007) | 0.71 (1.73e-09) |
| Single modal (CSF) | 0.73 (3.47e-20) | 0.78 (1.40e-05) | 0.72 (3.90e-16) |
| Single modal (MRI) | 0.70 (7.29e-22) | 0.73 (1.03e-12) | 0.69 (2.07e-16) |
| (c) 18 month prediction | |||
| Proposed | 0.79 | 0.82 | 0.79 |
| Baseline | 0.78 (0.08) | 0.81 (0.64) | 0.77 (0.15) |
| Single modal (cognitive performance) | 0.74 (2.43e-06) | 0.76 (0.03) | 0.71 (7.72e-08) |
| Single modal (CSF) | 0.72 (9.84e-12) | 0.78 (0.07) | 0.71 (6.55e-11) |
| Single modal (MRI) | 0.72 (5.26e-14) | 0.76 (0.16e-04) | 0.71 (2.20e-10) |
| (d) 24 month prediction | |||
| Proposed | 0.80 | 0.81 | 0.80 |
| Baseline | 0.76 (2.12e-05) | 0.80 (0.57) | 0.76 (8.95e-05) |
| Single modal (cognitive performance) | 0.74 (4.69e-07) | 0.76 (0.15) | 0.72 (1.20e-09) |
| Single modal (CSF) | 0.73 (1.06e-09) | 0.73 (0.001) | 0.74 (3.16e-07) |
| Single modal (MRI) | 0.71 (1.18e-12) | 0.76 (0.03) | 0.70 (2.53e-10) |