Table 2 Results using the original data with short sentences removed.
From: Multimodal deep learning for dementia classification using text and audio
Shorts-removed | Accuracy | Precision | Recall | F1-score | AUROC |
|---|---|---|---|---|---|
Audio | 0.5958 ± 0.014 | 0.587 ± 0.046 | 0.2945 ± 0.108 | 0.3811 ± 0.084 | 0.624 ± 0.022 |
Text | 0.6861 ± 0.027 | 0.623 ± 0.042 | 0.6951 ± 0.058 | 0.6545 ± 0.022 | 0.7593 ± 0.024 |
Audio \(+\) Time | 0.5897 ± 0.009 | 0.5631 ± 0.034 | 0.3554 ± 0.07 | 0.4306 ± 0.054 | 0.6163 ± 0.01 |
Text \(+\) Time | 0.6683 ± 0.029 | 0.6197 ± 0.043 | 0.7011 ± 0.137 | 0.6494 ± 0.032 | 0.7353 ± 0.047 |
Audio \(+\) Text | 0.6052 ± 0.014 | 0.5669 ± 0.024 | 0.5255 ± 0.152 | 0.534 ± 0.082 | 0.6482 ± 0.014 |
Audio \(+\) Text \(+\) Time | 0.6257 ± 0.065 | 0.5724 ± 0.075 | 0.5239 ± 0.151 | 0.5415 ± 0.107 | 0.6769 ± 0.081 |
