Engaging Machine Learning and Data Linkage to Understand Firearm Suicide Among Females

Overview

This study will use the National Violent Death Reporting System (NVDRS) and the Washington Violent Death Reporting System (WA-VDRS) to develop and implement a Natural Language Processing (NLP) approach to better understand and contextualize female firearm suicide. Using an NLP-enhanced WA-VDRS linked to multiple state-level administrative datasets, we will examine demographic and health care utilization patterns between cases.

Status

Completed

Purpose

To provide evidence to inform the development of tailored policies and interventions to reduce female firearm suicide. Our work and our partnership with agencies across the state of Washington will move the field of firearm suicide prevention forward.

Approach

We will first generate an enhanced NVDRS dataset through the creation of an NLP pipeline. The algorithms will then be used to code additional variables in the WA-VDRS, which will allow for a more detailed analysis of female firearm suicide. We will use the WA-VDRS NLP-enhanced data to create mutually exclusive typologies. We will then link the WA-VDRS to several sources of administrative claims data to examine health care utilization patterns and demographics across groups. We will also assess changes in demographic variables across typologies over the last 5 years to understand how demographics have shifted along the same period in which female firearm suicide rates were increasing.

Significance

Female firearm suicide has risen by 34% over the decade of 2008-2018. Prior studies show that risk factors for suicides that involve firearms versus other means are different and may require a unique and tailored approach to prevention. This work will categorize female firearm suicide and identify health care utilization patterns prior to the suicide to inform tailored interventions and identify settings for intervention delivery.

Published Research

• Characterizing Female Firearm Suicide Circumstances: A Natural Language Processing and Machine Learning Approach

  View research (subscription required) View paper on OSF

• Circumstantial Variables Preceding Firearm Suicide Among Females With and Without Mobility Disability in the USA: Comparative Analysis Using Data From the National Violent Death Reporting System

  View research

• Developing a Traditional Natural Language Pipeline (NLP) with National Violent Death Reporting System (NVDRS) Data

  Free online course

• Female Firearm Suicide Across the Lifespan: A Descriptive Assessment of Cases From the National Violent Death Reporting System

  View research (subscription required)

• Identifying Rare Circumstances Preceding Female Firearm Suicides: Validating A Large Language Model Approach

  View research

Investigator Bio

Laura Prater, Ph.D., MPH, MHA, is a research scientist in the Firearm Injury and Policy Research Program in the Harborview Injury Prevention and Research Center at the University of Washington. Dr. Prater is a health services researcher with a program of research focused on the understanding and prevention of firearm suicide. Dr. Prater's work focuses on developing and testing evidence-based interventions in both the policy and clinical context to ultimately end firearm suicide.