A Suicide Intervention App for College Students

As an AIM Clinical Science Fellow, Benjamin will use a smartphone application to predict suicide behavior in college students. 

Suicide is the second most common cause of death among young adults and suicide rates among this cohort has increased in recent years. Unfortunately, although a number of interventions have been developed to reduce suicide and self-harm among these individuals, the success of such interventions has proven to be limited. Improving prevention of young adult suicide may depend on improving prediction of suicidality.

Recent methodological developments have improved assessment and prediction of self-harm and suicide. First, the use of ecological momentary assessment (EMA) in daily life has enhanced assessment of a range of behavioral and psychological constructs. A number of recent EMA studies have successfully found links between momentary/daily negative affect, hopelessness, suicidal ideation, and other constructs and suicide and self-harm behavior.

Second, recent advances in machine learning have also enhanced prediction of suicidal behavior. Machine learning approaches such as random-forest regression and regularized regression outperform traditional predictive modeling in predicting suicide and associated constructs. However, there is a dearth of research combining machine learning algorithms with EMA to predict and intervene on suicidal behavior.

Dr. Johnson’s research study combines two aims:
1) apply machine learning to EMA data to predict the emergence of self-harm urges and suicidality among young adults
2) employ a mobile intervention to reduce the likelihood of such behaviors.

The study will not only improve on current predictive models of young adult suicidal behavior, but also provide initial data on mobile, disseminatable, and scalable prevention efforts for young adult suicidality.