A recent study conducted by UCLA Health researchers has found that machine learning models can be effective in preventing suicide among young people, in stark contrast to the shocking case of a Belgian man who took his own life after being encouraged to do so by an AI chatbot. Currently, healthcare providers rely on a data system to assess the mental health of children, but the study reveals that machine learning models are far superior in detecting self-injury thoughts or behaviors in children. Suicide is a leading cause of death among young people in Europe, with an estimated nine million children between the ages of 10 and 19 living with mental disorders. Anxiety and depression account for over half of all cases. In the US, the Department of Health and Human Services estimates that 20 million young people currently have a diagnosed mental health disorder. To evaluate the effectiveness of the current systems used to assess mental health, UCLA Health researchers reviewed clinical notes from 600 emergency department visits made by children aged 10 to 17. The findings were concerning, as the clinical notes failed to identify 29% of children with self-injurious thoughts or behaviors, and the statements made by health specialists overlooked 54% of patients. This was largely due to the fact that children often do not report suicidal thoughts and behaviors during their initial visit to the emergency department.

Even when the two systems were used together, 22% of at-risk children were still missed. The study also revealed that boys were more likely to be overlooked than girls, and Black and Latino youth were more likely to be left out compared to white children. However, the introduction of machine-learning models made a significant difference. The researchers created three models that analyzed various data, including previous medical care, medications, patient location, and lab test results, to estimate thoughts of suicide or self-injury. All three models outperformed the traditional methods in identifying at-risk children. Lead author of the study, Juliet Edgcomb, emphasized the importance of improving detection before jumping to prediction. While the machine-learning models did have a higher chance of false positives, Edgcomb stated that it is preferable to err on the side of caution and identify more children at risk. For individuals contemplating suicide, it is crucial to seek help and support. Befrienders Worldwide is an international organization with helplines available in 32 countries. If you find yourself in need, please reach out to them by visiting befrienders. org to find the appropriate helpline for your location.