Repository logo
 

Understanding Schizophrenia: A Bayesian Symptom Network Approach

Files

Thesis(4.6 MB)

Date

2025

Authors

Supervisor

Siegert, Richard John
Narayanan, Ajit
Sandham, Margaret
Vignes, Matthieu

Item type

Thesis

Degree name

Doctor of Philosophy

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

Schizophrenia is one of the most debilitating mental health conditions, marked by moderate to severe cognitive and functional impairments. With an increase in research on the aetiology, symptoms, and prognosis, researchers are close to developing robust social, biological, and cognitive conceptualisations of schizophrenia. A recent model of health is the network theory of mental disorders (NTMD), where symptoms of a mental disorder are modelled in a complex system. Here, symptoms cause or are dependent on the expression of other symptoms, named symptom networks (SN). The research in this thesis aims to review the SN research of schizophrenia, conduct an SN that evaluates the role of cognition, among other variables, and provide evidence for the NTMD. The first study in this thesis is a systematic review of SNs, which aims to identify major themes across studies and identify gaps in the literature. The findings in the systematic review highlight that cognition and functioning were central nodes across the included studies. Hallucinations and/or delusions were not central in most of the networks. This finding aligns with other evidence that proposes that schizophrenia is a disorder of cognition. Following the results of the systematic review, the second study aims to identify the role of cognition, sociodemographic, psychopathology, and quality of life (QOL) variables from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study using a Bayesian network (BN) framework. Findings indicate that processing speed significantly predicts all other cognitive processes and QOL, highlighting processing speed as a core cognitive impairment in schizophrenia. The third and final study aims to determine whether latent variable models (LVMs) and SNs can be integrated to improve model fit. It was found that the integrated model significantly improves the fit of an LVM alone when using the Positive and Negative Syndrome Scale (PANSS) from the CATIE study and the Scale of Psychosis-Risk Symptoms (SOPS) from the North American Prodromal Longitudinal Study 3 (NAPLS3) study. The results provide partial support for the complexity principle of the NTMD, which states that complex relationships between symptoms represent mental disorders. The collection of these three studies fills a gap in research using SNs to understand schizophrenia. By focusing on systems of symptoms, this thesis provides evidence for an alternative modelling approach to schizophrenia where complex interactions of symptoms are represented using SNs. Two integrative proposals extend from the results of this thesis: that processing speed is the prime cognitive impairment in schizophrenia and that the SN methodology may also have applications for use in clinical profiles for people living with mental illness in general.

Description

Keywords

Source

DOI

Publisher's version

Rights statement

Permanent link

http://hdl.handle.net/10292/20080

Collections

Doctoral Theses