Automatic depression score estimation with word embedding models
作者：
Highlights：
• A novel approach for using neural language models to estimate depression severity.
• The proposed approach uses different models depending on the symptom nature.
• Our classification method addresses the joint task using three variants.
• In a standard evaluation benchmark, our proposal outperforms prior solutions.
• We included a symptom-by-symptom analysis to study the performance of the variants.
摘要
•A novel approach for using neural language models to estimate depression severity.•The proposed approach uses different models depending on the symptom nature.•Our classification method addresses the joint task using three variants.•In a standard evaluation benchmark, our proposal outperforms prior solutions.•We included a symptom-by-symptom analysis to study the performance of the variants.
论文关键词：Depression prediction,Neural language models,Social media,Word embeddings
论文评审过程：Received 11 June 2021, Revised 8 June 2022, Accepted 18 August 2022, Available online 24 August 2022, Version of Record 31 August 2022.
论文官网地址：https://doi.org/10.1016/j.artmed.2022.102380
