How machine learning can identify gun buyers at risk of suicide

Suicide is a leading cause of death in the United States, with data showing one person takes their own life every 11 minutes.

Suicides also account for the majority of gun deaths in the country, and firearms are used in most completed suicides as about 90 percent of suicide attempts with a gun are fatal. 

Previous studies have also found an increased risk of suicide in the time period immediately following a handgun purchase, suggesting acquisition is a key risk factor.

Now, new research out of the University of California, Davis, suggests machine learning can forecast gun purchasers’ likelihood of firearm suicide through the use of handgun purchasing data. Identifying those at risk allows for prevention interventions and can ultimately help reduce suicide rates.

The proof-of-concept study was conducted by researchers at the Violence Prevention Research Program and is the first of its kind to investigate such a correlation. 

“While limiting access to firearms among individuals at increased risk for suicide presents a critical opportunity to save lives, accurately identifying those at risk remains a key challenge,” said study co-author Hannah S. Laqueur of the University of California in a statement. 

“Our results suggest the potential utility of handgun records in identifying high-risk individuals to aid suicide prevention.”

America is changing faster than ever! Add Changing America to your Facebook or Twitter feed to stay on top of the news.

Investigators assessed data from the California’s Dealer’s Record of Sale (DROS) database, which includes information on nearly 5 million handgun transaction records. Using random forest (RF) classification, they tested data’s ability to predict those at a heightened suicide risk within one year of purchase and identified which factors help determine this likelihood.

The purchases were completed by nearly 2 million individuals between 1996 and 2015, the majority of whom were men. California death record data were also employed to identify firearm suicides throughout the study period. 

Researchers found a 0.07 percent rate of suicide within one year of purchase (3,278 transactions among 2,614 individuals). Forty-one risk factors for firearm suicide were identified, including older age, being a first-time purchaser, white race, living in close proximity to the seller and purchasing of a revolver. 

“The type of firearm purchased was the most important predictor of firearm suicide,” reserachers wrote, adding month of purchase was also an important factor, as peaks were seen throughout the spring and early summer.

Of the 5 percent of purchases deemed most risky, nearly 40 percent were linked with a purchaser who died by firearm suicide. Of the transactions with a predicted probability score of 0.95 or greater based on the the RF classification, 69 percent were associated with a purchaser who committed suicide via firearm within one year. 

“This study contributes to the growing evidence that computational methods can aid in the identification of high-risk groups and the development of targeted interventions,” Laqueur said.

However, because many firearm suicides also occurred among low-risk individuals, additional interventional methods are needed. 

More research is also needed to improve the predictive performance of the algorithm developed. Adding in factors like presence of a mental health disorder, which was not included in the dataset employed, could improve risk predictions.

Efforts to better equip firearms dealers and sellers to identify those at risk of suicide have been underway. For example, the Gun Shop Project develops and distributed tip sheets, brochures, and other suicide prevention communications to retailers and range owners throughout the country. 

changing america copyright