Introduction
Since 2005, female firearm suicide rates increased by 34%, outpacing the rise in male
firearm suicide rates over the same period. The objective of this study was to develop
and evaluate a natural language processing pipeline to identify a select set of common
and important circumstances preceding female firearm suicide from coroner/medical
examiner and law enforcement narratives.
Methods
Unstructured information from coroner/medical examiner and law enforcement narratives
were manually coded for 1,462 randomly selected cases from the National Violent Death
Reporting System. Decedents were included from 40 states and Puerto Rico from 2014
to 2018. Naive Bayes, Random Forest, Support Vector Machine, and Gradient Boosting
classifier models were tuned using 5-fold cross-validation. Model performance was
assessed using sensitivity, specificity, positive predictive value, F1, and other
metrics. Analyses were conducted from February to November 2022.
Results
The natural language processing pipeline performed well in identifying recent interpersonal
disputes, problems with intimate partners, acute/chronic pain, and intimate partners
and immediate family at the scene. For example, the Support Vector Machine model had
a mean of 98.1% specificity and 90.5% positive predictive value in classifying a recent
interpersonal dispute before suicide. The Gradient Boosting model had a mean of 98.7%
specificity and 93.2% positive predictive value in classifying a recent interpersonal
dispute before suicide.
Conclusions
This study developed a natural language processing pipeline to classify 5 female firearm
suicide antecedents using narrative reports from the National Violent Death Reporting
System, which may improve the examination of these circumstances. Practitioners and
researchers should weigh the efficiency of natural language processing pipeline development
against conventional text mining and manual review.
To read this article in full you will need to make a payment
Purchase one-time access:
Academic & Personal: 24 hour online accessCorporate R&D Professionals: 24 hour online accessOne-time access price info
- For academic or personal research use, select 'Academic and Personal'
- For corporate R&D use, select 'Corporate R&D Professionals'
Subscribe:
Subscribe to American Journal of Preventive MedicineAlready a print subscriber? Claim online access
Already an online subscriber? Sign in
Register: Create an account
Institutional Access: Sign in to ScienceDirect
REFERENCES
- Web-Based injury statistics query and reporting system (WISQARS).Centers for Disease Control and Prevention, U.S.2021 (Updated December 2, 2021. Accessed February 16, 2023.)
- The relationship between firearms and suicide: a review of the literature.Aggress Violent Behav. 1999; 4: 59-75https://doi.org/10.1016/S1359-1789(97)00057-8
- Mental illness and reduction of gun violence and suicide: bringing epidemiologic research to policy.Ann Epidemiol. 2015; 25: 366-376https://doi.org/10.1016/j.annepidem.2014.03.004
- Results from the 2014 National Survey on Drug Use and Health: detailed tables.Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality, Rockville, MD2015 (Accessed August 16, 2022)
- Prior suicide attempts are less common in suicide decedents who died by firearms relative to those who died by other means.J Affect Disord. 2016; 189: 106-109https://doi.org/10.1016/j.jad.2015.09.007
- Lethality of firearms relative to other suicide methods: a population based study.J Epidemiol Community Health. 2003; 57: 120-124https://doi.org/10.1136/jech.57.2.120
- The epidemiology of firearm suicide in the United States.J Urban Health. 2002; 79: 39-48https://doi.org/10.1093/jurban/79.1.39
- Characteristics of adult male and female firearm suicide decedents: findings from the National Violent Death Reporting System.Inj Prev. 2009; 15: 322-327https://doi.org/10.1136/ip.2008.021162
- Firearm suicide among veterans in the general population: findings from the national violent death reporting system.J Trauma. 2009; 67: 503-507https://doi.org/10.1097/TA.0b013e3181b36521
- Changes in overall and firearm veteran suicide rates by gender, 2001–2010.Am J Prev Med. 2015; 48: 360-364https://doi.org/10.1016/j.amepre.2014.10.013
National Violent Death Reporting System (NVDRS). Centers for Disease Control and Prevention. https://www.cdc.gov/violenceprevention/datasources/nvdrs/index.html. Updated September 28, 2021. Accessed February 1, 2022.
NVDRS data access. Centers for Disease Control and Prevention. https://www.cdc.gov/violenceprevention/datasources/nvdrs/dataaccess.html. Updated September 28, 2021. Accessed October 1, 2022.
National Violent Death Reporting System (NVDRS) coding manual. Centers for Disease Control and Prevention. https://www.cdc.gov/violenceprevention/pdf/nvdrs/nvdrsCodingManual.pdf. Updated April 22, 2021. Accessed October 1, 2022.
- Job-related problems prior to nurse suicide, 2003–2017: a mixed methods analysis using natural language processing and thematic analysis.J Nurs Regul. 2021; 12: 28-39https://doi.org/10.1016/S2155-8256(21)00017-X
- An overview of content analysis.Pract Assess Res Eval. 2000; 7: 17https://doi.org/10.7275/z6fm-2e34
- Reliability and inter-rater reliability in qualitative research.Proc ACM Hum Comput Interact. 2019; 3: 1-23https://doi.org/10.1145/3359174
Dedoose Version 8.0.35. Dedoose. https://www.dedoose.com/. Accessed February 1, 2022.
- Multidimensional diagnostic criteria for chronic pain: introduction to the ACTTION–American Pain Society Pain Taxonomy (AAPT).J Pain. 2016; 17 (suppl): T1-T9https://doi.org/10.1016/j.jpain.2016.02.010
- Natural Language ToolKit (NLTK) Book. O’Reilly Media, Sebastopol, CA2009
Richardson L. Beautiful soup documentation. Read the Docs.2016. https://beautiful-soup-4.readthedocs.io/en/latest/index.html?highlight=Beautiful%20soup%20documentation.
- Scikit-learn: machine learning in Python.J Mach Learn Res. 2011; 12: 2825-2830
- Suicide among older adults living in or transitioning to residential long-term care, 2003 to 2015.JAMA Netw Open. 2019; 2e195627https://doi.org/10.1001/jamanetworkopen.2019.5627
Ko TM, Kalesnikava VA, Jurgens D, Mezuk B. A data science approach to estimating the frequency of driving cessation associated suicide in the U.S.: evidence from the national violent death reporting system. Front Public Health. 2021;9(1):689967. https://doi.org/10.3389/fpubh.2021.689967.
- The random forest algorithm for statistical learning.The Stata Journal. 2020; 20: 3-29https://doi.org/10.1177/1536867X20909688
- Random forests.Mach Learn. 2001; 45: 5-32https://doi.org/10.1023/A:1010933404324
- An Introduction to Statistical Learning - With Applications in R.Springer, Cham, Switzerland2013
- Selecting a classification method by cross-validation.Mach Learn. 1993; 13: 135-143https://doi.org/10.1007/BF00993106
- Nurse suicide in the United States: analysis of the Center for Disease Control 2014 National Violent Death Reporting System dataset.Arch Psychiatr Nurs. 2019; 33: 16-21https://doi.org/10.1016/j.apnu.2019.04.006
- Risk factors for child death during an intimate partner homicide: a case-control study.Child Maltreat. 2021; 26: 356-362https://doi.org/10.1177/1077559520983901
- Characteristics and patterns of older adult homicides in the United States.Inj Epidemiol. 2021; 8: 5https://doi.org/10.1186/s40621-021-00299-w
Mintz M, Bills S, Snow R, Jurafsky D. Distant supervision for relation extraction without labeled data. 2009;2:1003–1011. https://doi.org/10.3115/1690219.1690287.
- Named entity recognition without labelled data: a weak supervision approach.in: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. 2020https://doi.org/10.18653/v1/2020.acl-main.139
- Snorkel: Rapid Training Data Creation with Weak Supervision.Proceedings VLDB Endowment. 2017; 11: 269-282https://doi.org/10.14778/3157794.3157797
Brown TB, Mann B, Ryder N, et al. Language models are few-shot learners. ArXiv. Preprint. Online May 28, 2020. https://arxiv.org/abs/2005.14165. Accessed November 7, 2022.
- Machine learning and cloud-based knowledge graphs to recognize suicidal mental tendencies.Comput Intell Neurosci. 2022; 20223604113https://doi.org/10.1155/2022/3604113
Cerel J, Maple M, van de Venne J, Moore M, Flaherty C, Brown M. Exposure to suicide in the community: prevalence and correlates in one U.S. State. Public Health Rep. 2016;131(1):100–107. https://doi.org/10.1177/003335491613100116.
- Intimate partner violence, suicide, and their overlapping risk in women veterans: a review of the literature.Mil Med. 2019; 184: e201-e210https://doi.org/10.1093/milmed/usy355
- Intimate partner violence and incident depressive symptoms and suicide attempts: a systematic review of longitudinal studies.PLoS Med. 2013; 10e1001439https://doi.org/10.1371/journal.pmed.1001439
- Chronic pain among suicide decedents, 2003 to 2014: findings from the national violent death reporting system.Ann Intern Med. 2018; 169: 448-455https://doi.org/10.7326/M18-0830
- The influence of patient sex, provider sex, and sexist attitudes on pain treatment decisions.J Pain. 2014; 15: 551-559https://doi.org/10.1016/j.jpain.2014.02.003
- Using natural language processing to improve suicide classification requires consideration of race.Suicide Life Threat Behav. 2022; 52: 782-791https://doi.org/10.1111/sltb.12862
Article info
Publication history
Published online: March 16, 2023
Publication stage
In Press Journal Pre-ProofIdentification
Copyright
© 2023 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.
