Publikationsserver
Toward Disentangling the Heterogeneity of Phenotypes: Multivariate Statistical Approaches in Neuroimaging for Psychiatric Phenotyping
The biopsychosocial model offers a conceptual theoretical framework to explain mental health and illness. It states, that the continua of biological, psychological, and social factors must be seen in interaction in order to determine an individual’s risk for mental illness. By defining psychiatric p...
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Beteiligte: | |
| Format: | Dissertation |
| Sprache: | Englisch |
| Veröffentlicht: |
Philipps-Universität Marburg
2023
|
| Schlagworte: | |
| Online Zugang: | PDF-Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Zusammenfassung: | The biopsychosocial model offers a conceptual theoretical framework to explain mental health and illness. It states, that the continua of biological, psychological, and social factors must be seen in interaction in order to determine an individual’s risk for mental illness. By defining psychiatric phenotypes this model is transferred into practical application.
Advances in neuroimaging techniques hold big promises to contribute to psychiatric phenotyping. Being able to unravel associations between behavior and the brain, enables the detection of further biomarkers for psychiatric disorders besides genetics. However, considering the complexity of both the brain as well as mental disorders, mapping reliable brain-behavior relationships is challenging. So far, mostly univariate statistical approaches are applied in psychiatric neuroimaging research which do not fully live up to this complexity.
This dissertation applied multivariate statistical approaches to both brain as well as behavioral data to investigate the feasibility of those approaches in neuroimaging research for psychiatric phenotyping. The aim was to go one step further toward disentangling the heterogeneity of phenotypes in psychiatric neuroimaging.
In STUDY I we used the multivariate approach of Structural Equation Modeling (SEM) to build a comprehensive model of brain as well as clinical data in a large sample of patients with major depressive disorder (MDD). A previous published clinical SEM, which included risk, symptom, and cognitive variables was first replicated and then extended by a brain structural connectivity measurement. Findings of this study reflect on our understanding of white matter integrity in MDD and bring new insights into the relationship between an established risk factor as well as a core symptom of MDD with brain structural connectivity.
The data driven approach in STUDY II in form of cluster analysis aimed to explore the underlying biological grouping in a large transdiagnostic cohort. Results show that data driven subgroups based on a brain morphometric parameter do not align with the clinical diagnostic grouping. Results of subsequent correlational analyses with early environmental risk factors as well as neuropsychological variables hint toward a transdiagnostic involvement of this brain morphometric parameter in psychopathology rather than being bound to clinical diagnostic categories.
In STUDY III, results of Principal Component Analyses showed that conceptualizations as well as assessment of the personality trait schizotypy differ across studies which leads to a source of heterogeneity when this phenotype is used in neuroimaging studies. Confirmatory Factor Analyses provided a newly approach for the assessment of schizotypy.
In conclusion, this dissertation provided novel insights into the feasibility of multivariate statistical approaches in both psychiatric neuroimaging as well as psychometric research. Results of the studies highlight the importance of going beyond simple diagnostic borders to define reliable phenotypes in psychiatry. Future research in this field needs to shift toward more comprehensive approaches to capture the complexity of mental disorders. By this, a reliable foundation for computational models is built to enable the practical application of individual predictions about mental health and illness. |
|---|---|
| DOI: | 10.17192/z2024.0014 |