Learning effective machine learning models for clinical applications in psychiatry

• Author / Creator

  Sawalha, Jeffrey

• This is a comprehensive examination of the diagnostic landscape in psychiatry and the role precision psychiatry might play in redefining how we classify illnesses. This thesis delves into three areas that currently exist in psychiatry today and is divided into five chapters. The first and second chapters review the literature and clinical space of AI and psychiatry. The last three chapters consist of academic articles that explore the use of AI in the three problematic areas of psychiatry.The third chapter involves the early stage detection of Bipolar Disorder (Type 1) using cognitive assessments. Identifying cognitive dysfunction in the early stages of Bipolar Disorder (BD) can allow for early intervention. Previous studies have shown a strong correlation between cognitive dysfunction and the number of manic episodes. The objective of this study was to apply machine learning (ML) techniques on a battery of cognitive tests to identify first-episode BD patients (FE-BD). Specifically, we wanted to know if we could make generalized predictions about the various stages of BD using cognitive tests.The fourth chapter examines childhood anxiety and produces a learned model that can detect dysfunction in the brains of children while they examine emotional facial expressions. Childhood anxiety is a difficult disorder to diagnose due to validity controversies and the conflation of normal developmental-behavioral patterns with anxiety symptoms. Our study not only seeks to train a model that can distinguish anxious from non-anxious children, but also to discover neural markers related to this diagnosis.Lastly, the fifth chapter utilizes natural language processing to detect the presence of post-traumatic stress disorder (PTSD). Specifically, we utilize sentiment analysis, a sub area of natural language processing (NLP), to extract emotional content from text information. In our study, we train an ML model on text data, which is part of the Audio/Visual Emotion Challenge and Workshop (AVEC-19) corpus, to identify individuals with PTSD using sentiment analysis from semi-structured interviews. We sought to understand the emotional spectrum of language and compare our findings with the ongoing literature.Together, each of these studies illustrate how ML can be used to augment clinical decision-making surrounding the underlying conditions of individuals who may suffer from these illnesses. In doing so, we provide a conceptual review of the current barriers that exist in precision psychiatry today. Our hope is to provide the reader with a foundation of how ML can be used in psychiatry, while also highlighting some of the current barriers that hold back this field today. This comes in the form of a conceptual review (Chapter 2) and sets the landscape for the three published articles included.

• Subjects / Keywords

  • Psychiatry
  • Machine Learning
  • Computer Science
  • Translational Psychiatry
  • Neuroscience

• Graduation date

  Fall 2023

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-d2wm-sw02

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.