Understanding Suicidal Behaviours in the Context of Depression through a Biopsychosocial, Whole Person Modelling, Machine Learning Framework

Abstract

Suicide is a multifaceted psychiatric phenomenon with various biopsychosocial drivers. Suicide can be further categorized into suicidal ideation and self-harm, with self-harm encompassing both fatal and non-fatal suicide attempts. There is ongoing debate as to the merit of differentiating nonsuicidal self-injury from other suicidal behaviours (such as suicide attempts). This differentiation is commonly attributed to the difference in the underlying intent of the action. Empirical studies based on ideation-to-action theories of suicidal behaviour have had little success in the identification of specific yet stable risk factors for suicide. Generally, suicidal behaviours are used as diagnostic criteria for other psychiatric disorders—namely major depressive disorder, bipolar disorder, and borderline personality disorder. Furthermore, the interplay between suicidal thoughts and behaviours and the genetics of depression is not well understood. Overall, there is a movement away from traditional risk prediction frameworks and toward suicide resilience as a novel research concept to frame future studies. In order to address these research gaps, three cross-sectional studies were conducted using data from the Toronto Adolescent and Youth Cohort study, the Canadian Longitudinal Study on Aging, and the UK Biobank. First, the data-driven, operationalized constructs of self-harm based on intent (i.e., into distinct categories of suicidal behaviour and nonsuicidal self-injury) were shown to be clinically differentiable in a population of treatment seeking, transitional-aged youth. Second, suicidal thoughts but not behaviours were shown to mediate associations between genetic risk for depression and peripheral biomarkers (specifically white blood cell count, neutrophil count, and triglyceride levels). Lastly, machine learning models developed within a resilience-based, Whole Person Modelling framework prioritized age, age at first sexual intercourse, and educational attainment as top predictors of suicide attempt resilience. These studies further our understanding of suicidal behaviours and the genetic underpinning of depression. Collectively, they highlight the biopsychosocial factors of resilience to suicide attempts in the context of depression and provide candidate targets for future studies. Focused individual-level interventions guided by these population-level findings may help decrease rates of suicidal behaviours by bolstering resilience, particularly in relation to depression.