Table 4 Evaluation results of the DAIC-WOZ dataset

From: Depression detection methods based on multimodal fusion of voice and text

Modality

Metric

MCN

Bi-LSTM

MCN+Bi-LSTM

Audio

Accuracy

0.9434

0.9496

–

Precision

0.8603

0.9608

–

Recall

0.9277

0.9157

–

F1 score

0.8928

0.9021

–

Text

Accuracy

0.7021

0.7471

–

Precision

0.6506

0.6723

–

Recall

0.7020

0.7470

–

F1 score

0.6753

0.7077

–

Audio+Text

Accuracy

0.9343

0.9274

0.9794

Precision

0.8170

0.7926

0.9702

Recall

0.9548

0.9669

0.9631

F1 score

0.8806

0.8711

0.9666

  1. Significant values are in bold.
Back to article page