Table 3 Results using the augmented versions of the original data.
From: Multimodal deep learning for dementia classification using text and audio
Original-augmented | Accuracy | Precision | Recall | F1-score | AUROC |
|---|---|---|---|---|---|
Audio | 0.6038 ± 0.036 | 0.5846 ± 0.021 | 0.1831 ± 0.04 | 0.2764 ± 0.044 | 0.6336 ± 0.003 |
Text | 0.8294 ± 0.005 | 0.8339 ± 0.022 | 0.751 ± 0.042 | 0.7892 ± 0.013 | 0.9208 ± 0.002 |
Audio \(+\) Time | 0.6306 ± 0.068 | 0.5994 ± 0.012 | 0.3673 ± 0.044 | 0.4542 ± 0.033 | 0.6759 ± 0.008 |
Text \(+\) Time | 0.8344 ± 0.006 | 0.8267 ± 0.035 | 0.7721 ± 0.043 | 0.797 ± 0.009 | 0.9236 ± 0.005 |
Audio \(+\) Text | 0.825 ± 0.013 | 0.7978 ± 0.039 | 0.7859 ± 0.056 | 0.7899 ± 0.019 | 0.9124 ± 0.015 |
Audio \(+\) Text \(+\) Time | 0.8315 ± 0.014 | 0.8212 ± 0.034 | 0.767 ± 0.012 | 0.7927 ± 0.014 | 0.9177 ± 0.014 |
