JMIR Publications

ABSTRACT

Background:

Suicide is one of the leading causes of death among young and middle-aged people. However, little is understood about the behaviors leading up to actual suicide attempts and as well as whether these behaviors are specific to the nature of suicide attempts.

Objective:

The goal of this study was to examine the clusters of behaviors taking place antecedent to suicide attempt to determine if they could be used to assess potential lethality of the attempt. To accomplish this goal, we developed a deep learning model using the relationships between antecedent to suicide attempts and the attempt itself.

Methods:

This study used data from the Korea National Suicide Survey. We identified 1112 individuals who attempted suicide and completed a psychiatric evaluation in the emergency room. The 15 items Beck’s Suicide Intent Scale (SIS) was used for assessing antecedent behaviors and the medical outcomes of the suicide attempts was measured by assessing lethality with Columbia Suicide Severity Rating Scale, C-SSRS (lethal suicide attempts >3; non-lethal attempt 3).

Results:

Using scores from the SIS, lethal and non-lethal attempters comprised two different networks nodes with the edges representing the relationships between nodes. Among the antecedent behaviors, having the conception of method’s lethality predicted suicidal behaviors with severe medical outcomes. The vectorized relationship values between the elements of antecedent behaviors in our deep learning model (E-GONet) increased the precision in identifying lethal attempts by up to 6%, compared with other models (mean precision: E-GONet = 0.84, linear regression = 0.78, random forest = 0.82).

Conclusions:

The relationships between behaviors antecedent to suicide attempts can be used to understand the suicidal intent of individuals and help identify the lethality of a potential suicide attempt. Such models may be useful in prioritizing cases for preventive intervention.