Table 2. Classification rates.
| Approach | Dim. red.method | Accuracy | Sensitivity | Specificity | Positive | Negative |
| Likelihood | Likelihood | |||||
| PCA | 73.91% | 76.92% | 70.00% | 2.56 | 0.33 | |
| Only images | PLS | 78.26% | 84.62% | 70.00% | 2.82 | 0.22 |
| ICA | 69.57% | 76.92% | 60.00% | 1.92 | 0.38 | |
| Only psych. scores | – | 73.91% | 73.08% | 75.00% | 2.92 | 0.36 |
| Psych. sc.+MMSE+Age | – | 84.78% | 84.62% | 85.00% | 5.64 | 0.18 |
| Images and | PCA | 80.43% | 84.62% | 75.00% | 3.38 | 0.21 |
| psych. scores | PLS | 84.78% | 88.46% | 80.00% | 4.42 | 0.14 |
| (Early integration) | ICA | 76.09% | 80.77% | 70.00% | 2.69 | 0.27 |
| Images and psych. | PCA | 80.43% | 84.62% | 75.00% | 3.38 | 0.21 |
| scores+MMSE+Age | PLS | 84.78% | 88.46% | 80.00% | 4.42 | 0.14 |
| (Early integration) | ICA | 73.91% | 76.92% | 70.00% | 2.56 | 0.33 |
| Images and | PCA | 82.61% | 92.31% | 70.00% | 3.08 | 0.11 |
| psych. scores | PLS | 84.78% | 88.46% | 80.00% | 4.42 | 0.14 |
| (Interm. integration) | ICA | 82.61% | 84.62% | 80.00% | 4.23 | 0.19 |
| Images and psych. | PCA | 89.13% | 92.31% | 85.00% | 6.15 | 0.09 |
| scores+MMSE+Age | PLS | 89.13% | 92.31% | 85.00% | 6.15 | 0.09 |
| (Interm. integration) | ICA | 89.13% | 92.31% | 85.00% | 6.15 | 0.09 |
| Images and | PCA | 86.96% | 92.31% | 80.00% | 4.62 | 0.10 |
| psych. scores | PLS | 84.78% | 88.46% | 80.00% | 4.42 | 0.14 |
| (Late integration) | ICA | 80.43% | 80.77% | 80.00% | 4.04 | 0.24 |
| Images and psych. | PCA | 89.13% | 92.31% | 85.00% | 6.15 | 0.09 |
| scores+MMSE+Age | PLS | 89.13% | 92.31% | 85.00% | 6.15 | 0.09 |
| (Late integration) | ICA | 89.13% | 92.31% | 85.00% | 6.15 | 0.09 |
Accuracy, sensitivity, specificity and positive and negative likelihoods for the systems implemented. These rates were estimated by means of a leave-one-out cross-validation scheme and using the database described above.