Table 3 Above: dementia scoring results for DenseNet-121 and LLE. Bold denotes the highest result. Below: dementia scoring with off-diagonal acceptance for DenseNet-121 and LLE. The measures show very accurate results.
From: Automatic dementia screening and scoring by applying deep learning on clock-drawing tests
Class | Precision | Recall | f1-score | AUC |
|---|---|---|---|---|
Score 1 | 0.67 | 0.79 | 0.73 | 0.84 |
Score 2 | 0.67 | 0.51 | 0.58 | 0.72 |
Score 3 | 0.84 | 0.91 | 0.87 | 0.93 |
Score 4 | 0.76 | 0.72 | 0.74 | 0.84 |
Score 5 | 0.64 | 0.63 | 0.63 | 0.80 |
Score 6 | 0.74 | 0.78 | 0.76 | 0.88 |
Class | Precision | Recall | f1-score | AUC |
|---|---|---|---|---|
Score 1 | 1.00 | 1.00 | 1.00 | 1.00 |
Score 2 | 1.00 | 0.98 | 0.99 | 0.99 |
Score 3 | 1.00 | 0.99 | 0.99 | 1.00 |
Score 4 | 0.99 | 0.94 | 0.96 | 0.97 |
Score 5 | 1.00 | 1.00 | 1.00 | 1.00 |
Score 6 | 0.93 | 1.00 | 0.96 | 0.99 |
