Reaching Those at Highest Risk for Suicide: Development of a Model Using Machine Learning Methods for use With Native American Communities

Research output: Contribution to journalArticle

Abstract

Objective: Suicide prevention is a major priority in Native American communities. We used machine learning with community-based suicide surveillance data to better identify those most at risk. Method: This study leverages data from the Celebrating Life program operated by the White Mountain Apache Tribe in Arizona and in partnership with Johns Hopkins University. We examined N = 2,390 individuals with a validated suicide-related event between 2006 and 2017. Predictors included 73 variables (e.g., demographics, educational history, past mental health, and substance use). The outcome was suicide attempt 6, 12, and 24 months after an initial event. We tested four algorithmic approaches using cross-validation. Results: Area under the curves ranged from AUC = 0.81 (95% CI ± 0.08) for the decision tree classifiers to AUC = 0.87 (95% CI ± 0.04) for the ridge regression, results that were considerably higher than a past suicide attempt (AUC = 0.57; 95% CI ± 0.08). Selecting a cutoff value based on risk concentration plots yielded 0.88 sensitivity, 0.72 specificity, and a positive predictive value of 0.12 for detecting an attempt 24 months postindex event. Conclusion: These models substantially improved our ability to determine who was most at risk in this community. Further work is needed including developing clinical guidance and external validation.

Original languageEnglish (US)
JournalSuicide and Life-Threatening Behavior
DOIs
StateAccepted/In press - Jan 1 2019

Fingerprint

North American Indians
Suicide
Area Under Curve
Decision Trees
Aptitude
Population Groups
Mental Health
History
Demography
Machine Learning
Sensitivity and Specificity

ASJC Scopus subject areas

  • Clinical Psychology
  • Public Health, Environmental and Occupational Health
  • Psychiatry and Mental health

Cite this

@article{9601411a4ddc4d66b178d0d9b95bdd50,
title = "Reaching Those at Highest Risk for Suicide: Development of a Model Using Machine Learning Methods for use With Native American Communities",
abstract = "Objective: Suicide prevention is a major priority in Native American communities. We used machine learning with community-based suicide surveillance data to better identify those most at risk. Method: This study leverages data from the Celebrating Life program operated by the White Mountain Apache Tribe in Arizona and in partnership with Johns Hopkins University. We examined N = 2,390 individuals with a validated suicide-related event between 2006 and 2017. Predictors included 73 variables (e.g., demographics, educational history, past mental health, and substance use). The outcome was suicide attempt 6, 12, and 24 months after an initial event. We tested four algorithmic approaches using cross-validation. Results: Area under the curves ranged from AUC = 0.81 (95{\%} CI ± 0.08) for the decision tree classifiers to AUC = 0.87 (95{\%} CI ± 0.04) for the ridge regression, results that were considerably higher than a past suicide attempt (AUC = 0.57; 95{\%} CI ± 0.08). Selecting a cutoff value based on risk concentration plots yielded 0.88 sensitivity, 0.72 specificity, and a positive predictive value of 0.12 for detecting an attempt 24 months postindex event. Conclusion: These models substantially improved our ability to determine who was most at risk in this community. Further work is needed including developing clinical guidance and external validation.",
author = "Haroz, {Emily E.} and Walsh, {Colin G.} and Novalene Goklish and Cwik, {Mary F.} and Victoria O’Keefe and Allison Barlow",
year = "2019",
month = "1",
day = "1",
doi = "10.1111/sltb.12598",
language = "English (US)",
journal = "Suicide and Life-Threatening Behavior",
issn = "0363-0234",
publisher = "Wiley-Blackwell",

}

TY - JOUR

T1 - Reaching Those at Highest Risk for Suicide

T2 - Development of a Model Using Machine Learning Methods for use With Native American Communities

AU - Haroz, Emily E.

AU - Walsh, Colin G.

AU - Goklish, Novalene

AU - Cwik, Mary F.

AU - O’Keefe, Victoria

AU - Barlow, Allison

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Objective: Suicide prevention is a major priority in Native American communities. We used machine learning with community-based suicide surveillance data to better identify those most at risk. Method: This study leverages data from the Celebrating Life program operated by the White Mountain Apache Tribe in Arizona and in partnership with Johns Hopkins University. We examined N = 2,390 individuals with a validated suicide-related event between 2006 and 2017. Predictors included 73 variables (e.g., demographics, educational history, past mental health, and substance use). The outcome was suicide attempt 6, 12, and 24 months after an initial event. We tested four algorithmic approaches using cross-validation. Results: Area under the curves ranged from AUC = 0.81 (95% CI ± 0.08) for the decision tree classifiers to AUC = 0.87 (95% CI ± 0.04) for the ridge regression, results that were considerably higher than a past suicide attempt (AUC = 0.57; 95% CI ± 0.08). Selecting a cutoff value based on risk concentration plots yielded 0.88 sensitivity, 0.72 specificity, and a positive predictive value of 0.12 for detecting an attempt 24 months postindex event. Conclusion: These models substantially improved our ability to determine who was most at risk in this community. Further work is needed including developing clinical guidance and external validation.

AB - Objective: Suicide prevention is a major priority in Native American communities. We used machine learning with community-based suicide surveillance data to better identify those most at risk. Method: This study leverages data from the Celebrating Life program operated by the White Mountain Apache Tribe in Arizona and in partnership with Johns Hopkins University. We examined N = 2,390 individuals with a validated suicide-related event between 2006 and 2017. Predictors included 73 variables (e.g., demographics, educational history, past mental health, and substance use). The outcome was suicide attempt 6, 12, and 24 months after an initial event. We tested four algorithmic approaches using cross-validation. Results: Area under the curves ranged from AUC = 0.81 (95% CI ± 0.08) for the decision tree classifiers to AUC = 0.87 (95% CI ± 0.04) for the ridge regression, results that were considerably higher than a past suicide attempt (AUC = 0.57; 95% CI ± 0.08). Selecting a cutoff value based on risk concentration plots yielded 0.88 sensitivity, 0.72 specificity, and a positive predictive value of 0.12 for detecting an attempt 24 months postindex event. Conclusion: These models substantially improved our ability to determine who was most at risk in this community. Further work is needed including developing clinical guidance and external validation.

UR - http://www.scopus.com/inward/record.url?scp=85074791875&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074791875&partnerID=8YFLogxK

U2 - 10.1111/sltb.12598

DO - 10.1111/sltb.12598

M3 - Article

C2 - 31692064

AN - SCOPUS:85074791875

JO - Suicide and Life-Threatening Behavior

JF - Suicide and Life-Threatening Behavior

SN - 0363-0234

ER -