Identifying Preliminary Risk Profiles for Dissociation in 16- to 25-year-olds Using Machine Learning

Introduction: Dissociation is associated with clinical severity, increased risk of suicide and self-harm, and disproportionately affects adolescents and young adults. Whilst evidence indicates multiple factors contribute to dissociative experiences, a multi-factorial explanation of increased risk for dissociation has yet to be achieved.
Methods: We used multiple regression to investigate the relative influence of five plausible risk factors (childhood trauma, loneliness, marginalisation, socio-economic status, and everyday stress), and machine learning to generate tentative high-risk profiles for ‘felt sense of anomaly’ dissociation (FSA-dissociation) using cross-sectional online survey data from 2384 UK-based 16- to 25-year-olds.
Results: Multiple regression indicated that four risk factors significantly contributed to FSA-dissociation, with relative order of contribution: everyday stress, childhood trauma, loneliness, and marginalisation. Exploratory analysis using machine learning suggested dissociation results from a complex interplay between interpersonal, contextual, and intrapersonal pressures: alongside marginalisation and childhood trauma, negative self-concept and depression were important in younger (16-20 years), and anxiety and maladaptive emotion regulation in older (21-25 years) respondents.
Conclusions: Validation of these findings could inform clinical assessment, and prevention and outreach efforts, improving the under-recognition of dissociation in mainstream services.