What is it about?
Suicide is a widespread public health concern. It is the tenth leading cause of death and even with increased study over the past fifty years, the suicide rate has risen nearly every year over the last twenty. Theories of suicide have been proposed to understand the causes of suicide. Likewise, predictive computational models have been deployed to identify patients who may go on to attempt or die by suicide. In the present project we compared the accuracy between many common theories of suicide and predictive models. Theories of suicide focused on predicting suicide from risk factors based in hopelessness, emotion dysregulation, loss of connection and feeling like a burden on loved ones, as well as biological vulnerabilities. Predictive models generally employed dozens of risk factors in large, healthcare wide data sets. Findings indicated that in the prediction of suicidal ideation, suicide attempts, and suicide deaths, predictive models were significantly more accurate than theories of suicide. While predictive models were more accurate in prediction of suicide, theories of suicide may give us more insight and understanding of the causes of suicide.
Photo by Albert Bleeker on Unsplash
Why is it important?
This work is important in that it highlights the benefits of both approaches in suicide prediction. With theories of suicide, researchers and clinicians gain understanding in the way that risk factors may lead to the development of suicide. This highlights potential treatment and intervention targets. With predictive models, classification of future suicide related outcomes are prioritized. This highlights detection of patients who may be at risk and would benefit from treatment or intervention programs.
Read the Original
This page is a summary of: A direct comparison of theory-driven and machine learning prediction of suicide: A meta-analysis, PLoS ONE, April 2021, PLOS, DOI: 10.1371/journal.pone.0249833.
You can read the full text:
The following have contributed to this page