Hispanic Firearm Suicide Decedents: Identifying Circumstances and Typologies with National Violent Death Reporting System Data

Nearly 1 in 2 Hispanic adult suicide deaths involves a firearm, the most lethal suicide method. However, there are critical gaps in characterizing Hispanic firearm suicide deaths and identifying opportunities for preventing firearm suicide among Hispanic adults. Our objectives are to leverage the power of machine

learning (ML) to identify circumstances preceding firearm suicides among Hispanic adults and generate new information on the typology of Hispanic adult firearm suicide decedents. We propose to achieve these objectives through the following Specific Aims: (1) Develop a natural language processing (NLP) pipeline to

identify and code circumstances preceding firearm suicide deaths among Hispanic adults and (2) Establish mutually-exclusive clinical typologies of Hispanic adult firearm suicide decedents with distinct combinations of demographic characteristics and circumstances preceding death. This project will ultimately provide critical

information that will be used to inform targeted intervention and evaluation opportunities through a future R01, such as implementing lethal means counseling, advising when and for which patients the intervention should be provided. The proposed research is significant for two reasons. First, this project will fill a critical need for

identifying key circumstances preceding highly-lethal firearm suicide deaths among Hispanic adults. Since the circumstances preceding firearm suicide death can differ by race/ethnicity, we will use NIMHD’s Health Disparities Research Framework to organize how we categorize the circumstances at the individual level.

Second, this project will generate mutually-exclusive clinical typologies of Hispanic adult firearm suicide decedents, including insights into the distinct demographic characteristics and circumstances experienced by different subgroups of Hispanic adult firearm suicide decedents. The proposed research is innovative for three reasons. First, this will be among the first studies to

leverage free-text information abstracted from coroner/medical examiner (CME) and law enforcement (LE) reports to generate much-needed information on the circumstances preceding Hispanic firearm suicides. Second, this will be the first project to leverage the power of NLP and ML with free-text CME and LE data to

predict common circumstances preceding firearm suicide death among Hispanics quickly and accurately while also creating standalone, annotated NLP tools for other users and applications. Third, this will be the first study funded to establish distinct subgroups of Hispanic firearm suicide decedents and their unique combinations of

demographic and circumstance characteristics.