1. Background
Repetitive negative thought (RNT) patterns, such as worry and rumination, are common and distressing symptoms that are characteristic of mood and anxiety disorders, but present across diagnostic categories. However, despite the obvious clinical importance of these symptoms, they are typically assessed using simple self-report measures. As a result, we have a poor understanding of how they manifest dynamically in the real world, and what psychological and physiological characteristics are predictive of everyday RNT. Better measurement of real-world RNT could drive improvements in diagnostics and treatment. In particular, it will be important to build models that can capture and model dynamic trajectories within timeseries data of symptoms and concurrent psychological and physiological processes.
2. Novelty & importance
Some prior research has used ecological momentary assessment (EMA) methods to assess real-world RNT. However, there has been limited application of powerful data-driven approaches to cluster and characterise dynamic RNT trajectories and tie them to cognitive and physiological characteristics. This project will fill this gap by capturing real-world measures of RNT through EMA, in conjunction with cognitive assessments and physiological measures. Ultimately, this project provide models that bring together different modalities to understand how RNT evolves in daily life, and how cognitive and physiological factors influence its dynamics.
3. Aims & objectives
The overarching objective of this project is to apply machine learning methods to characterise and predict real-world RNT trajectories. In particular, the project will focus on building models that link symptom timeseries data from EMA together with data from other modalities (e.g., cognitive assessment, physiology).
Specific aims are to:
1) Apply timeseries clustering methods to identify RNT trajectory types within EMA data
2) Identify correspondence between RNT trajectories and cognitive processing measured synchronously using gamified, remote assessments
3) Build multimodal (e.g., psychological, physiological, neural) predictive models of RNT trajectories