How A San Diego Researcher Aims To Predict Military Suicides
Thursday, November 13, 2014
Photo by Staff Sgt. Justin Phemister / U.S. Army
Researchers say a computer model they've built could help identify Army soldiers most likely to commit suicide.
Researchers say a predictive computer model they've built could help identify soldiers in the United States Army most likely to commit suicide.
The researchers developed the model by pouring through data on more than 40,000 soldiers who'd been hospitalized for mental health problems. They found these patients were more likely to commit suicide than most army personnel.
But they also found patterns suggesting a small group within this subpopulation is disproportionately driving recent spikes in military suicides.
"The neat thing about these big data, machine learning methods is that you can put them all together and come up with a profile," said UC San Diego School of Medicine's Murray Stein, co-author of the new study published in JAMA Psychiatry.
"You get a pretty good idea about an individual and where they are in this risk profile."
Stein and his colleagues found that among soldiers recently discharged from psychiatric hospitals, more than half of suicides were committed by just five percent of patients. Those most at risk included men who enlisted later in life and committed crimes during service.
Stein said this model could be used to flag medical records and alert doctors about patients who fit the profile for heightened suicide risk.
"While they're still in hospital, we can identify them and provide some specialized care or some additional follow-up for this very high-risk group."
To view PDF documents, Download Acrobat Reader.