A Yale study published as part of a series on gun violence has garnered national media attention for its treatment of firearm violence as a contagious social epidemic.

By visualizing instances of gun violence within social networks, researchers were able to predict and account for nearly two-thirds of Chicago’s arrests for gun violence over an eight-year period.

“We wanted to model gunshot violence as a contagious disease that spreads from individual to individual based on their social interactions and their social network, as opposed to just something that happens randomly or spreads from neighborhood to neighborhood over space,” first author Ben Green ’14 said.

To accomplish this, Green devised a machine-learning mathematical model that created social interaction networks based on the criminal arrest data of 138,246 individuals. Two people become part of the same network if they were arrested together, and the researchers hypothesized that gun violence would “cascade” through a network in the same way a contagious disease is transmitted through a community.

According to the study, the team’s model was able to predict 63.1 percent of gunshot violence episodes in Chicago between 2006 and 2014, a higher percentage than models based on demographics or social contagion alone could. Additionally, the team found evidence to support a cascading model — on average, 125 days separated two shootings within the same co-offending network.

This information is crucial for pinpointing high-risk individuals who might benefit from intervention, said Andrew Papachristos, a corresponding author of the study and professor of sociology at Yale.

“If we have this social map, we can send first responders, trauma specialists, interventionists and police if necessary,” he said.

Papachristos said he initially considered the study as a way to take the analogy of gun violence as an epidemic beyond just “a political punchline.”

This study is the first to model gun violence as a contagion over a large-scale social network, Green said. Previous studies have instead looked at aggregate risk factors, such as being poor, living in a particular neighborhood, being young or being a member of a gang, Papachristos added.

A commentary and editorial accompanied the study, which was published Jan. 3 in JAMA Internal Medicine as part of the journal’s ongoing series on firearm injuries and gun violence, Editor-at-Large Robert Steinbrook said.

A concurrently published study done by researchers at the University of Pennsylvania looked at the association between drug and alcohol use and adolescent firearm homicides in Philadelphia.

Steinbrook said he wrote the journal’s editorial to place both the Yale study and the Penn study within the context of the larger issue of gun violence.

JAMA Internal Medicine’s series on firearm injuries and gun violence began in 2016, when Steinbrook co-authored an editorial calling for papers. Even though gun violence is neither a health condition nor an infection, “It’s a major cause of suffering and death, particularly in urban areas, as well as a major public health issue for us, and we wanted to call attention to it,” Steinbrook said.

Charlie Branas, a co-author of both the Penn study and the commentary on the Yale study, said his group’s commentary was in part prompted by a desire to underscore the success of an epidemiological approach to model gun violence.

“We think that this signals very clearly that gun violence is a disease problem and something that the public health and medical communities deserve to have a say in, and resources should be expended studying it as if it were cancer or heart disease or Ebola,” he said.

The commentary also questions the Yale researchers’ decision to spurn a spatial model and focus only on people when trying to prevent gun violence, Branas said.

According to Steinbrook’s editorial, a potential weakness of the study is its lack of information on respondents’ history of “substance abuse, employment, kinship, and gang membership.”

“Plenty of other studies have done a better job bisecting the different risk factors,” Papachristos said. “Because we used official data, we don’t have a lot of the fine-grain data.”

Although the arrest record information the team used also contained biases “in terms of who gets arrested,” they were still able to build a comprehensive network from it, Green said.

Major media outlets, both scientific and lay, have picked up the Yale team’s research. Papachristos said he was happy with the publicity the study was getting; however, some articles “go straight for the predictive policing angle and ignore the fact that what we’re trying to do is the opposite,” he said.

Green said he hopes the study will shift intervention strategies away from policing-based efforts towards public health-based efforts. He added that he hopes it will also move the conversation away from viewing individuals as offenders towards viewing them as victims. In a public health epidemic, you save the lives of individuals because they matter, Papachristos said, giving the example of treating an obese person, regardless of their body mass index.

“We should be doing the same thing with gun violence,” he said. “We should treat these individuals as victims and afford them the same care we would of people who are involved in any other epidemic.”

The study provides evidence in support of fewer, focused criminal justice contacts within a community with the potential to save lives. However, there remains the core problem of how to prevent an individual from entering into high-risk networks in the first place, Papachristos said.

Papachristos added that he is currently working on another study, applying the original paper’s methodology to multiple other American cities. Although the analysis is still in its early stages, the epidemiological principle of cascading seems to hold, he said.

In 2014, firearm homicides accounted for 10,945 deaths — nearly 70 percent of all homicides — according to the Centers for Disease Control and Prevention.