The effect of sleep-wake behaviors on the onset of mania in youth: a computational agent-based model.

Bipolar disorder (BD) is highly disabling and often leads to premature mortality. Interventions focus on treating acute episodes, maintaining euthymia or preventing recurrence. A clinical priority is preventing transition from euthymia or depression (normal or low activation states) to mania (a high activation state). Identifying causal mechanisms underlying the temporal dynamics of this transition may enable more effective early intervention and secondary prevention. We demonstrate how agent-based modelling aids evaluating causal hypotheses for the onset and course of mania in youth by quantifying how perturbations in sleep-wake and circadian mechanisms precipitate transitions to high activation (manic) states. The proposed framework can be used as a digital laboratory to interrogate hypotheses about causal mechanisms that give rise to mania and the potential impacts of various behavioural interventions. A longitudinal simulation study of digital agents is performed, using the research evidence base, input from clinicians, and youth with lived experience. This is done using a purposely created digital twin model of youth individuals, which operationalizes the relationships between individual 24-hour cycles of sleep-wake behaviors (SWB) and the onset of manic episodes. A population of 200 digital ‘agents’ with individual characteristics derived from young people at different stages of BD from the longitudinal study of the Brain and Mind Youth Cohort (N=2,330) is studied. Multiple simulation scenarios that explore and quantify how typical features of the 24-hour cycle of SWB and the underlying circadian system impact the temporal dynamics of mania are investigated.