Advertisement

Study: Predictive analysis can ID soldiers at risk for violent crime

A model developed using data from all members of the U.S. Army over the course of a decade accurately predicted a significant number of violent crimes.

By Stephen Feller
Although the model was accurate for a high number of crimes, researchers caution the lack of evidence that intense intervention can be beneficial. File photo by Keizo Mori/UPI
Although the model was accurate for a high number of crimes, researchers caution the lack of evidence that intense intervention can be beneficial. File photo by Keizo Mori/UPI | License Photo

BOSTON, Oct. 6 (UPI) -- A predictive model that uses administrative data for members of the U.S. Army was shown to accurately predict whether soldiers would later commit violent crime, according to a new study by Harvard Medical School.

The researchers caution that while the model appeared to predict with some accuracy, there is no evidence to support high-risk preventive interventions being mounted by the Army.

Advertisement

The research at Harvard was done in collaboration with Army STARRS, a large study conducted by the Army from 2009 to 2015 on mental health risk and resilience among its members.

"These numbers are striking," said Ronald Kessler, the McNeil Family Professor of Health Care Policy at Harvard Medical School, in a press release. "They show us that predictive analytic models can pinpoint the soldiers at highest violence risk for preventive interventions. Targeting such interventions might be the best way to bring down the violent crime rate in the Army."

Researchers at Harvard created an actuarial model using machine learning methods to analyze data from a consolidated administrative database for the 975,057 members of the U.S. Army as part of Army STARRS between 2004 and 2009. The model was validated using an independent sample collected between 2011 and 2013.

Advertisement

During the five-year period, 5771 soldiers committed a violent crime, such as murder-manslaughter, kidnapping, aggravated arson, aggravated assault or robbery. The researchers found key indicators of potential violent crime included indicators of disadvantaged social or socioeconomic status, early career stage, prior crime, and mental disorder treatment.

The model showed the 5 percent of soldiers most at risk between 2004 and 2009 accounted for 36.2 percent of physical violent crimes committed by men and 33.1 percent of those committed by women. From 2011 to 2013, the 5 percent with highest predicted risk were responsible for 50.5 percent of the major violent crimes.

"It is important to recognize that severe violent crimes are uncommon even in this high-risk group," said John Monahan, a professor of law at the University of Virginia. "This means that implementing intensive high-risk preventive interventions would make sense only if the interventions are shown to be highly efficient -- something that has not yet been demonstrated."

The study is published in the journal Psychological Medicine.

Latest Headlines