Identifying psychiatric manifestations in outpatients with depression and anxiety: a large language model-based approach
Authors:
Open Access
- 2 December 2025
- journal article
- research article
- Published by Springer Nature in npj Mental Health Research
- Vol. 4 (1) , 63
Abstract
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.Keywords
Funding Information
- Tianqiao and Chrissy Chen Institute (2023-TX-018, 2023-TX-018, 2023-TX-018, 2023-TX-018, 2023-TX-018, 2023-TX-018)
- National Natural Science Foundation of China (82071500, 82071500, 82071500, 82071500, 82071500, 82071500)
- Program of Shanghai Academic/Technology Research Leader (21XD1423300, 21XD1423300, 21XD1423300, 21XD1423300, 21XD1423300, 21XD1423300)
This publication has 47 references indexed in Scilit:
- Large Language Model Influence on Diagnostic ReasoningJAMA Network Open, 2024
- Early Prediction and Detection of Anxiety Level Using Support Vector Machine2023
- Review and content analysis of textual expressions as a marker for depressive and anxiety disorders (DAD) detection using machine learningDiscover Artificial Intelligence, 2023
- Artificial intelligence assisted tools for the detection of anxiety and depression leading to suicidal ideation in adolescents: a reviewCognitive Neurodynamics, 2022
- Predictive utility of symptom measures in classifying anxiety and depression: A machine-learning approachPsychiatry Research, 2022
- Automated assessment of psychiatric disorders using speech: A systematic reviewLaryngoscope Investigative Otolaryngology, 2020
- TF–IDF2011
- Symptom Checklist‐90‐Revised2010
- The Hamilton Rating Scale for Depression1986
- A Self-Rating Depression ScaleArchives of General Psychiatry, 1965