AI model better at foreseeing soldier suicide attempts than Army health evaluation, study says

Army Considers AI to Boost Suicide Prevention

The U.S. Army is exploring a novel approach to strengthen its suicide prevention initiatives. A recent study, published in Nature Mental Health, examined if machine learning could help predict soldiers at a higher risk of self-harm.

Researchers developed an artificial intelligence model using Army records from 2014 to 2019, analyzing 452,473 active-duty soldiers and 2,096 recorded suicide attempts. The model incorporated data from 54 Army and Defense Department systems, going beyond traditional risk questions.

The results were striking. The algorithm identified that 25% of soldiers in the highest predicted risk category accounted for 69.5% of suicide attempts within six months. This meant that the yearly Periodic Health Assessment, which screens for issues, was not as effective on its own.

Drilling down, 95% of suicide attempts occurred among soldiers who had denied suicidal thoughts during their health assessment, highlighting the importance of a more sophisticated predictive approach.

However, the study also unveiled the stigma surrounding mental health in the military. Despite Pentagon efforts, service members continue to worry about being stigmatized if they speak up about their struggles.

While the study didn’t examine the cost of interventions based on the model, experts believe the potential benefits may outweigh any additional expenses. The Army could use this tool to refer soldiers to behavioral health treatment and provide more intensive care for those already in treatment.

This initiative comes amidst the military’s ongoing struggle with suicide, recently brought to the fore by the New Year’s Day suicide of an Army Special Forces member, Master Sgt. Matthew Livelsberger. Despite displaying no concerning behaviors during leave approval, his tragic death underscores the need for proactive, innovative suicide prevention strategies.