UC San Diego Prof's Algorithm Aims to Prevent Military Suicides

A 5-year research study, conducted a UC San Diego professor, aims to find a way to prevent suicides among service members using computer algorithms.

The U.S. Army is funding the $75 million Study To Assess Risk and Resilience in Service members (STARRS), which it says is the largest study of mental health risk and resilience ever conducted among military personnel.

A component of the study looked at 50,000 members of the Army who were hospitalized and had a mental health diagnosis, but not necessarily from a psychiatric ward. Some of those studied had physical injuries, but others did not.

Researchers entered more than a million pieces of data, including medical notes, prescriptions, criminal reports and other information, into a computer program. 

"We basically put all of these in and let the computer algorithms pull out the combination that gave us the best predictor,” said Murray Stein, MD., M.P.H., a psychiatry professor and researcher with UC San Diego.

The suicide predictor was a combination of factors. For instance, male patients who had a recent criminal history may have acted violently and had some kind of psychosis.

“Things like violence, drunk and disorderly, carrying a weapon onto base that wasn’t authorized — those all seemed to increase risk,” said Stein.

He found 5 percent – or 2,500 – of the active-duty Army members studied had those predictors, and they accounted for more than half of all of the suicides carried out or completed during the 5-year study, according to Stein.

“So we were able to identify out of the 50,000, a pretty small group of individuals who are at very, very high risk of suicide,” he said.

That information, said Stein, will help the Army predict those at risk and help those soldiers.

Contact Us