Identifying psychiatric manifestations in outpatients with depression and anxiety: a large language model-based approach
Accurate psychiatric diagnosis and assessment are crucial for effective treatment. However, current diagnostic approaches heavily rely on subjective observations constrained by time and clinical resources. This study investigates the potential of using Large Language Models (LLMs) to identify the symptoms in psychiatrist-patient dialogues and use them as intermediate features to predict the diagnostic labels. We collected audio recordings of 1160 outpatients with depressive disorder and anxiety disorder. LLMs were trained and utilized to identify clinical symptoms, rate assessment scales, and an ensemble learning pipeline was designed to classify diagnostic results and symptoms with 10-fold cross-validation. The system achieved 86.9% accuracy for identifying the appearance of clinical annotations and 74.7% (77.2%) accuracy for identifying symptoms of anxiety (depression). In addition, analysis of LLM-generated features shows that depression cases exhibited prominent markers of anhedonia and decreased volition, whereas anxiety disorders were characterized by tension and an inability to relax.
Authors
- Hong Qiu
- Yong Wang (ORCID: https://orcid.org/0000-0002-0496-8700)
- Jianyin Qiu (ORCID: https://orcid.org/0000-0002-8075-2215)
- Shu Zhang (ORCID: https://orcid.org/0000-0002-0614-3550)
- Z. Wang
- Shihao Xu (ORCID: https://orcid.org/0000-0001-8295-5692)
- Yihui Ding (ORCID: https://orcid.org/0000-0001-6246-8897)
- Li Feng (ORCID: https://orcid.org/0000-0002-0536-1791)
- Nils Opel (ORCID: https://orcid.org/0000-0003-4749-3298)
- Shao Jun
- Lidong Bing (ORCID: https://orcid.org/0000-0003-4565-6313)
- Yu Mei (ORCID: https://orcid.org/0009-0008-3126-2974)
- Tao Yang (ORCID: https://orcid.org/0000-0002-0961-9542)
- Qing Hua Zhang (ORCID: https://orcid.org/0000-0002-6595-8995)
- Jianhua Chen (ORCID: https://orcid.org/0000-0002-2560-3188)
- Yiming Yan (ORCID: https://orcid.org/0009-0008-0391-1092)
- Wen Gu (ORCID: https://orcid.org/0000-0002-1615-6064)
- Haiyang Geng (ORCID: https://orcid.org/0000-0001-6115-807X)
- Min Zhao (ORCID: https://orcid.org/0000-0001-8382-3050)
- Yifeng Xu (ORCID: https://orcid.org/0000-0002-0403-1141)
- Xun Jiang (ORCID: https://orcid.org/0009-0007-0118-3311)
- Chao Bao Luo (ORCID: https://orcid.org/0000-0001-7863-0545)
- Haoyun Tang
- Yan Li (ORCID: https://orcid.org/0000-0002-1755-8823)
- Yunyun Han
- Hao Liu (ORCID: https://orcid.org/0000-0002-2378-195X)
Institutions
- Shanghai Jiao Tong University (CN)
- Shanghai Mental Health Center (CN)
- Jena University Hospital (DE)
- Community Health Center (US)
- Kantar Health (United States) (US)
- Shanghai Jian Qiao University (CN)
- Shanghai Clinical Research Center (CN)
- Genomic Health (United States) (US)
Publication Details
- Journal
- npj Mental Health Research
- Published
- 2025-12-02
- DOI
- https://doi.org/10.1038/s44184-025-00175-1
- Citations
- 1
- Primary Topic
- Mental Health via Writing
- Type
- article
- Field-Weighted Citation Impact
- 2.44
Funders
- Natural Science Foundation of Shanghai
- National Natural Science Foundation of China
- Program of Shanghai Academic Research Leader
- National Social Science Fund of China