Comparing emotion feature extraction approaches for predicting depression and anxiety
Hannah Burkhardt, Michael Pullmann, Thomas Hull, Patricia Areán, Trevor Cohen
Abstract
The increasing adoption of message-based behavioral therapy enables new approaches to assessing mental health using linguistic analysis of patient-generated text. Word counting approaches have demonstrated utility for linguistic feature extraction, but deep learning methods hold additional promise given recent advances in this area. We evaluated the utility of emotion features extracted using a BERT-based model in comparison to emotions extracted using word counts as predictors of symptom severity in a large set of messages from text-based therapy sessions involving over 6,500 unique patients, accompanied by data from repeatedly administered symptom scale measurements. BERT-based emotion features explained more variance in regression models of symptom severity, and improved predictive modeling of scale-derived diagnostic categories. However, LIWC categories that are not directly related to emotions provided valuable and complementary information for modeling of symptom severity, indicating a role for both approaches in inferring the mental states underlying patient-generated language.- Anthology ID:
- 2022.clpsych-1.9
- Volume:
- Proceedings of the Eighth Workshop on Computational Linguistics and Clinical Psychology
- Month:
- July
- Year:
- 2022
- Address:
- Seattle, USA
- Editors:
- Ayah Zirikly, Dana Atzil-Slonim, Maria Liakata, Steven Bedrick, Bart Desmet, Molly Ireland, Andrew Lee, Sean MacAvaney, Matthew Purver, Rebecca Resnik, Andrew Yates
- Venue:
- CLPsych
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 105–115
- Language:
- URL:
- https://aclanthology.org/2022.clpsych-1.9
- DOI:
- 10.18653/v1/2022.clpsych-1.9
- Cite (ACL):
- Hannah Burkhardt, Michael Pullmann, Thomas Hull, Patricia Areán, and Trevor Cohen. 2022. Comparing emotion feature extraction approaches for predicting depression and anxiety. In Proceedings of the Eighth Workshop on Computational Linguistics and Clinical Psychology, pages 105–115, Seattle, USA. Association for Computational Linguistics.
- Cite (Informal):
- Comparing emotion feature extraction approaches for predicting depression and anxiety (Burkhardt et al., CLPsych 2022)
- PDF:
- https://preview.aclanthology.org/ingest-acl-2023-videos/2022.clpsych-1.9.pdf
- Data
- GoEmotions