A Computational Approach to Understanding Early Trauma and Paediatric PTSD

Zhang, Yi (2022) A Computational Approach to Understanding Early Trauma and Paediatric PTSD. Doctoral thesis, University of East Anglia.

[thumbnail of Thesis_MS_YZ_100223810_Complete.pdf]
Preview
PDF
Download (3MB) | Preview

Abstract

Research on post-traumatic stress disorder (PTSD) has developed successful theories to understand the condition and to inform intervention. However, it faces challenges regarding efficiently explaining individual differences, comorbidity and certain treatments. The computational movement in mental health research not only opens up new methods to study the phenomena in PTSD but provides new prospects to looking at PTSD.

The computational approach is a data-centred research method. Using complex mathematical modelling, the approach seeks new ways to describe and record phenomena in mental functioning (i.e., phenotyping). With big data, the approach aims to discover more robust patterns (i.e., data mining), and ultimately, to build models (i.e., computational modelling) that improve explanatory power and predictive accuracy. Although the computational approach has potential, its application in clinical psychology lacks systematic understanding and theoretical guidance.

The thesis aims to explore the computational approach in child and adolescent PTSD research. Four empirical studies were conducted to: investigate symptomatic trajectory and PTSD-depression comorbidity; trauma memory and appraisal; chronic PTSD prediction and risk interpretation; and the long-term impact of early stress on panic disorder. Importantly, all the studies utilized unconventional computational methods, including trajectory modelling, natural language processing, supervised machine learning modelling, interpretable machine learning techniques, and robust variance estimation.

The four studies serve as successful implementations of computational methodologies. The advantages of these methods are explained in the context of the necessity for computational phenotyping and the benefits of data mining. The findings address PTSD-specific research questions, concluding that negative trauma appraisals, memory coherence, cognitive avoidance and physiological reactions are critical factors to PTSD symptom development, comorbidity and individual differences. A review of those findings provides the basis for an in-depth discussion of acute stress symptoms, pre-trauma factors, long-term impact and the omission of physiological aspect in the cognitive-behavioural model of PTSD. The thesis concludes by proposing a preliminary computational model of PTSD.

Item Type: Thesis (Doctoral)
Faculty \ School: Faculty of Medicine and Health Sciences > Norwich Medical School
Depositing User: Chris White
Date Deposited: 03 Oct 2022 11:45
Last Modified: 03 Oct 2022 11:45
URI: https://ueaeprints.uea.ac.uk/id/eprint/88805
DOI:

Actions (login required)

View Item View Item