Australia (Commonwealth Union) – Suicide is a global issue which is highlighted on World Suicide Prevention Day. Economically underprivileged individuals are usually classified as having a greater chance of suicide, which is a major concern for authorities working to prevent suicide, with the current economic climate of high inflation, a further strain with more severe consequences for low-income families.
Approximately nine Australians commit suicide every day and the main cause of death for Australians between 15–44 years old. Suicide attempts are more frequent, with some estimates showing there occurrence up to 30 times as frequent as deaths.
Researchers at the University of New South Wales (UNSW) are hopeful that an Artificial Intelligence (AI) algorithm could be a valuable tool against suicide. Karen Kusuma, a UNSW Sydney PhD candidate in psychiatry at the Black Dog Institute, who evaluates adolescent suicide prevention said: “Suicide has large effects when it happens. It impacts many people and has far-reaching consequences for family, friends and communities.”
For the study, Kusuma’s team from the Black Dog Institute and the Centre for Big Data Research in Health recently evaluated the evidence base of machine learning models and their capability to forecast suicidal behaviors and thoughts. They looked into the performance of 54 machine learning algorithms formed earlier by researchers to forecast suicidal outcomes of ideation and attempted deaths.
Meta-analysis discovered that machine learning models outperformed traditional risk prediction models in forecasting suicide-related outcomes, which have traditionally not done so well. Kusuma also said: “Overall, the findings show there is a preliminary but compelling evidence base that machine learning can be used to predict future suicide-related outcomes with very good performance.”
Applying machine learning algorithms to forecast suicide-related results is still an unfolding research area, with 80 % of the identified studies published in the last 5 years. According to Kusuma, future research will also assist in addressing the risk of aggregation bias seen in algorithmic models so far.
