Reproduce “Depression and Anxiety Prediction Using Deep Language Models and Transfer Learning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Tomasz Rutowski, Elizabeth Shriberg, Amir Harati, Yang Lu, Piotr Chlebek, Ricardo Oliveira
Researchers at Ellipsis Health Inc. developed deep language models to detect depression and anxiety from conversational speech, achieving AUC scores of 0.828 for depression and 0.792 for anxiety. Their models demonstrated improved performance, reaching AUCs of 0.861 for depression and 0.845 for anxiety, when predicting the joint presence or absence of both conditions.
Key insights
- Jointly predicting the presence or absence of both depression and anxiety yields higher accuracy than predicting each condition individually.
- Deep language models can effectively capture linguistic patterns in conversational speech relevant to detecting mental health conditions.
- Word sequence cues in speech appear to be more salient for detecting depression compared to anxiety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.