Machine learning models to predict adolescent suicide and self-harm risk

AI is increasingly being used in mental health to detect high-risk individuals.

Share

Suicide is the leading cause of mortality among teenagers in Australia, and 18% of those aged 14-17 self-harm. Both have become more common in this age group in the recent decade. When a young person enters a health care facility, such as a hospital, with suicidal or self-harming behavior, clinicians assess the risk of suicide and self-harm. Adolescents outside healthcare settings often go unnoticed, undermining risk assessment’s effectiveness.

Researchers from UNSW Sydney, the Ingham Institute for Applied Medical Research, and the South Western Sydney Local Health District (SWSLHD) have developed ML models to predict the risk of suicide and self-harm attempts in adolescents. These models were more accurate than a standard approach, with previous suicide and self-harm attempts as the only risk factor.

In mental health, artificial intelligence (AI) is increasingly used to detect at-risk individuals. Machine learning (ML) algorithms can analyze vast amounts of patient data, detecting possible risk factors and assessing their ability to predict mental health disorders such as suicide and self-harm attempts.

Senior author Dr. Daniel Lin, who is a psychiatrist and mental health researcher affiliated with UNSW, the Ingham Institute, and SWSLHD, said, “Sometimes we need to digest and process much information that would be beyond the ability of the clinician; that’s the reason we are tapping into machine learning algorithms.”

The researchers examined data from the Longitudinal Study of Australian Children, which has gathered data from children nationwide since 2004. Their study included 2809 participants, divided into two age groups: 14-15 years and 16-17 years. The information comes from questionnaires completed by the children, their caregivers, and their instructors. Among the 2809 participants, 10.5 percent reported an act of self-harm, and 5.2 percent reported attempting suicide at least once in the past 12 months.

Dr. Lin said, “These behaviors are under-reported, so the proportions are higher.” 

The researchers found over 4,000 possible risk variables from the data, including mental health, physical health, interactions with others, and the school and home environment. They used a random forest classification algorithm (a sophisticated machine learning approach) to determine which risk variables at 14-15 were most predictive of suicide and self-harm attempts at 16-17. 

They also found that depressed moods, emotional and behavioral issues, self-perceptions, and school and family relationships were the most relevant risk factors for suicide and self-harm.

The researcher said, “We found that the young person’s environment plays a bigger role than we thought. This is a good thing from the standpoint of prevention because we now know that there’s more we can do for these individuals.” 

He added, “Parental support and school support is very important… We need to figure out how, as a society, we can support parenting and school education to protect our younger generation.”

Further study is required before these ML models may be used in therapeutic settings. To determine if the algorithms are still helpful in predicting suicide and self-harm attempts, they must be applied to real-world clinical datasets. The researchers also want to know how different risk factors influence behavior.

The researcher said, “A unique predictor of suicide was lack of self-efficacy when someone feels a lack of control over their environment and future. And a unique predictor of self-harm was lack of emotional regulation.”

Journal Reference:

  1. Raymond Su, Ping-I Lin, et al. Machine learning-based prediction for self-harm and suicide attempts in adolescents. Psychiatry Research. DOI: 10.1016/j.psychres.2023.115446
Latest Updates

Trending