Suicide attempt risk predicts inconsistent self-reported suicide attempts: A machine learning approach using...

Area of Science:

• Psychiatry
• Adolescent Health
• Machine Learning in Healthcare

Background:

• Inconsistent self-reports of lifetime suicide attempts (LSAs) hinder accurate assessment of suicidal behavior.
• Adolescent suicidality is a significant public health concern requiring improved risk assessment tools.

Purpose of the Study:

• To investigate the relationship between suicide attempt risk and the consistency of self-reported LSAs in adolescents.
• To determine if higher-risk adolescents report LSAs more consistently than lower-risk adolescents.

Main Methods:

• A machine learning model was trained on baseline data from a longitudinal sample of Norwegian adolescents (N=10,739).
• The model estimated LSA risk scores, which were then correlated with the consistency of LSA reporting at a 2-year follow-up.

Main Results:

• Internalizing problems, optimism, conduct problems, substance use, and disordered eating were key factors in suicide attempt risk.
• Adolescents with consistently reported LSAs showed significantly higher baseline suicide attempt risk.
• Inconsistent reporting of LSAs was associated with being male, lower depression, and fewer conduct problems, potentially leading to false negatives in risk assessment.

Conclusions:

• Consistent self-reporting of LSAs may signal higher suicide attempt risk.
• Findings support the Theory of Adolescent Suicidality (TAS) and can enhance suicide risk assessment accuracy.
• Inconsistent self-reported LSAs appear to indicate lower suicide attempt risk.