Table 1 The comparison of existing depression detection datasets.
From: A Multimodal Depression Consultation Dataset of Speech and Text with HAMD-17 Assessments
Datasets | Language | Modality | #Inst. | #Persons | Length | Scale | Data Source |
|---|---|---|---|---|---|---|---|
AVEC201320 | German | video, audio | 150 | 292 | — | BDI-II | human-computer interaction |
AVEC201421 | German | video, audio | 300 | 84 | 274 min | BDI-II | human-computer interaction |
DAIC-WoZ16 | English | video, audio, text | 189 | 193 | 2,756 min | PHQ-8 | human-computer interaction |
E-DAIC35 | English | video, audio, text | 275 | 351 | 4,282 min | PHQ-8 | human-computer interaction |
BlackDog17 | English | video, audio, text | 60 | 60 | — | DSM-IV | answering open-ended questions |
Mundt36 | English | audio | 35 | 35 | — | HAMD-17, QIDS | automated telephone interface |
MODMA19 | Chinese | EEG, audio | 53 | 53 | 431 min | PHQ-9 | real-world clinical consultation |
DepressionEmo25 | English | text | 6,037 | — | — | Reddit posts | |
WU3D24 | Chinese | text | — | 30,000 | — | — | Weibo posts |
PDCH (Ours) | Chinese | audio, text | 100 | 100 | 2,937 min | HAMD-17 | real-world clinical consultation |
