Complex machine learning models with many variables (approx. 50 predictors) were found to more accurately predict short-term suicide-specific outcomes relative to conventional univariate approaches.

Longitudinal study using online recruitment of civilians at high risk of short-term or imminent suicidal behavior to examine novel risk factors over 3 days, 2 week, and 4 week follow-up to find indicators of imminent suicidal behavior and develop effective methods of risk assessment during imminent risk periods.