iSPEAK: a Natural Language Processing Study in youth with subclinical psychotic experiences

Introduction

Up to one fifth of young people report that they have experienced unusual perceptual experiences, thoughts, or beliefs. These experiences (which we call Psychotic Experiences or PE) can take the form of hearing voices, seeing visions, feeling paranoid, or believing things that are not true.

Research Background

As part of my ERC Consolidator Award (iHEAR), I and my team have learned a great deal more about these experiences and what they mean for young people. We know that these experiences can be transitory in most cases, but that in some young people, they recur throughout adolescence and adulthood.

Impact on Mental Health

We know that, for some young people, these experiences can be associated with poor functioning and poorer mental health, but for others, they are not pathological. We have identified the need for an instrument that can differentiate between pathological and non-pathological psychotic experiences. This will support early risk screening and the development of interventions among those who experience PE.

Research Proposal

We propose to search for natural language patterns within self-reported PE that are associated with high and low levels of mental ill-health. Our hypothesis is that young people's first-person accounts of their experiences are likely to contain natural language phrases that could be used to differentiate between:

1. PEs that occur in the context of mental ill-health and are likely to require intervention.
2. PEs that do not require intervention.

Significance of the Study

Knowledge of natural language patterns for PE would be a valuable step in improving assessment tools in clinical practice. Additionally, this could be a valuable new resource for digital tools and e-Health, which is a growing area for mental health.