• Analysis of language patterns...

*Result*: Analysis of language patterns in schizophrenia based on natural language processing.

Title:
Analysis of language patterns in schizophrenia based on natural language processing.
Authors:
Huang J; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Xu J; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Zhao Y; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Zhang M; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Tian Z; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Qu W; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Tan Y; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Wang Z; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China., Tan S; Beijing HuiLongGuan Hospital, Peking University HuiLongGuan Clinical Medical School, Beijing, 100096, China. Electronic address: shupingtan@126.com.
Source:
Psychiatry research [Psychiatry Res] 2026 Feb; Vol. 356, pp. 116894. Date of Electronic Publication: 2025 Dec 11.
Publication Type:
Journal Article
Language:
English
Journal Info:
Publisher: Elsevier/North-Holland Biomedical Press Country of Publication: Ireland NLM ID: 7911385 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1872-7123 (Electronic) Linking ISSN: 01651781 NLM ISO Abbreviation: Psychiatry Res Subsets: MEDLINE
Imprint Name(s):
Publication: Limerick : Elsevier/North-Holland Biomedical Press
Original Publication: Amsterdam, Elsevier/North-Holland Biomedical Press.
MeSH Terms:
Schizophrenia*/complications , Schizophrenia*/physiopathology , Schizophrenia*/diagnosis , Emotions*/physiology , Natural Language Processing*, Humans ; Male ; Female ; Adult ; Machine Learning ; Middle Aged ; Young Adult
Contributed Indexing:
Keywords: Language dysfunction; Natural language processing; Schizophrenia
Entry Date(s):
Date Created: 20251218 Date Completed: 20260102 Latest Revision: 20260102
Update Code:
20260130
DOI:
10.1016/j.psychres.2025.116894
PMID:
41411708
Database:
MEDLINE

*Further Information*

*Background: Schizophrenia is a severe mental disorder, with language dysfunction being one of its prominent features. Studies have shown that language expression patterns in individuals with schizophrenia differ significantly from those of healthy controls. However, emotional states may influence language expression. This study aims to use natural language processing techniques to analyze stable language expression differences in individuals with schizophrenia under various emotional stimuli.
Methods: The study recruited 104 patients diagnosed with schizophrenia and 80 healthy controls. Participants watched videos depicting different emotional stimuli and then responded to standardized questions. A two-way multivariate analysis of variance was used to explore group differences in language expression. Multivariate linear regression was applied to investigate the relationship between text features and symptom severity. Additionally, machine learning techniques were employed to preliminarily assess the potential of language features in assisting diagnosis.
Results: Stable and significant differences were observed between the schizophrenia group and the healthy control group in conjunction usage and the standard deviation of sentence length. Specific language features, such as the frequency of the term "Jiushi" (equivalent to "just" or "exactly" in English), number of words, and mean dependency distance, showed significant and consistent associations with the severity of negative symptoms. Machine learning models achieved a diagnostic accuracy of 66.3 %.
Conclusions: Language expression in individuals with schizophrenia is influenced by emotional states, which may obscure certain disease-related features. However, stable language expression differences, such as conjunction frequency and sentence length variability, remain. Furthermore, language features can serve as stable predictors for multiple dimensions of negative symptoms.
(Copyright © 2025 Elsevier B.V. All rights reserved.)*

*Declaration of competing interest The authors declare no competing interests.*