Machine Learning: Researchers can predict suicidal gun buyers

A US research team from the Violence Prevention Research Program (VPRP) at the University of California has shown in a study that machine learning can be used to identify which handgun buyers are suicidal. Accordingly, a risk of suicide can be recognized on the basis of individual and common characteristics.

According to the study “Machine Learning Analysis of Handgun Transactions to Predict Firearm Suicide Risk”, which was published in Jama Network Open, previous studies have already shown that the risk of suicide is particularly high shortly after purchasing a handgun in the USA. The researchers concluded that the purchase itself can be an indicator of an increased risk of suicide.

The science team examined a data set of almost 5 million firearm purchases that were available from the California Dealer Record of Sale database. The datasets from 1996 to 2015 included almost 2 million people. Additionally, the researchers correlated this data with that from California death registers between 1996 and 2016.

The research team was able to identify 41 predictor variables from the sales data that enable a prognosis about the risk of suicide among gun buyers. Examples of predictive variables include handgun category (e.g., revolver or semi-automatic pistol), caliber size, price, location of gun purchase (close to home of purchaser), gun history, sex, ethnicity, and age.

The scientists applied a random forest classification algorithm to the purchase data set in order to be able to make predictions about suicides with firearms within one year. The research found that of the 5 percent of sales that the algorithm identified as particularly risky, 40 percent were related to a buyer who killed himself with a gun within a year. When the prediction probability was 0.95 or higher, 69 percent of those buyers committed suicide with a firearm within a year.

Hannah S. Laqueur, assistant professor and lead author of the study, sees a clear and strong association between the possession and acquisition of firearms and the risk of firearm suicide. The study proves that computer-aided methods can identify high-risk groups and “help develop targeted interventions”. With the predictions, those at risk of suicide could be identified, identified and saved from suicide through prevention.

However, the science team concedes that there is no such thing as a 100% predictor. According to Laqueur, people whose risk the algorithm assessed as “low” also committed suicide. Other forms would have to be found here in order to be able to help these people preventively. The researchers therefore see their study more as a proof of concept.

Around 48,000 Americans died by suicide in 2020, according to a study by the Centers for Disease Control and Prevention. More than half of these were handgun suicides.

A notice: In Germany, those affected can find help with problems and suicidal thoughts at telefonseelsorge.de and by telephone on 0800 1110111. Free help is also available in Switzerland (for young people and children) and in Austria.


(olb)

To home page

#Machine #Learning #Researchers #predict #suicidal #gun #buyers

Leave a Comment

Your email address will not be published.