Markov model maps teen suicide risk: NSSI signals instability in 11,864 youth

A new paper from an international team (Lee, Cardoen, Townsend, O'Connor, et al.) applies a Markov framework to adolescent mental health data, modelling year-to-year transitions in suicidal ideation and behaviour. Using longitudinal self-reports from 11,864 children in the Adolescent Brain Cognitive Development (ABCD) Study, they defined eight states combining suicidal ideation/behaviour with the presence or absence of non-suicidal self-injury (NSSI). The time-inhomogeneous discrete-time Markov chain revealed structured but developmentally shifting transition probabilities between ages 9 and 13.

Key findings: a generally high probability of recovery to a no-report state following ideation or behaviour, but a distinct trajectory for children with co-reported lifetime NSSI. These children were more likely to transition to or persist in suicidal behaviour, less likely to recover, and showed greater uncertainty in their transition likelihoods. The authors argue that NSSI marks both elevated risk and trajectory instability, offering a scalable, interpretable method for analysing large, sparse mental health datasets. This framework could improve early monitoring and time-varying risk assessment in youth mental health.