Table 3 Evaluation results of the CMDC dataset

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

Modality

Metric

MCN

Bi-LSTM

MCN+Bi-LSTM

Audio

Accuracy

0.8636

0.9613

–

Precision

0.8751

0.9608

–

Recall

0.8772

0.9604

–

F1 score

0.8636

0.9605

–

Text

Accuracy

0.7604

0.7710

–

Precision

0.7577

0.7684

–

Recall

0.7534

0.7722

–

F1 score

0.7549

0.7691

–

Audio+Text

Accuracy

0.9680

0.9226

0.9747

Precision

0.9691

0.9206

0.9752

Recall

0.9658

0.9221

0.9664

F1 score

0.9673

0.9213

0.9708

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