Predicting Substance Use among Military Veterans with a Positive MST Screen: A Machine Learning Approach

通过积极的 MST 筛选来预测退伍军人的药物使用情况：一种机器学习方法

• 批准号: 10458304
• 负责人: SHANNON FORKUS
• 金额: $ 0.25万
• 依托单位: UNIVERSITY OF RHODE ISLAND
• 依托单位国家: 美国
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-05-24 至 2023-05-23
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10458304
• 关键词: Address; Affect; Algorithms; Area; Big Data; Cessation of life; Clinical; Code; Complex; Data; Data Set; Detection; Development; Distress; Etiology; Exposure to; Goals; Grant; Health; Health behavior; Impairment; Individual; Institution; Intervention; Investigation; Knowledge; Legal; Link; Machine Learning; Maintenance; Manuscripts; Maps; Methodology; Military Personnel; Mission; National Institute of Drug Abuse; Nature; Occupational; Outcome; Output; Overdose; Policies; Population; Population Research; Prevention; Prognosis; Psychology; Public Health; Recording of previous events; Research; Research Methodology; Research Personnel; Research Project Grants; Rhode Island; Risk; Risk Factors; Sampling; Science; Self Medication; Service provision; Sexual Harassment; Statistical Data Interpretation; Statistical Methods; Suicide; Symptoms; Time; To specify; Traffic accidents; Translating; United States Department of Veterans Affairs; Universities; Veterans; career; clinically relevant; cohort; design; experience; high risk; high risk behavior; high risk population; improved; innovation; longitudinal analysis; longitudinal dataset; machine learning method; military service; military veteran; prediction algorithm; predictive modeling; preventive intervention; programs; prospective; sexual assault; sexual trauma; skills; social; stem; substance misuse; substance use; supervised learning; theories; trauma exposure

Project Summary/Abstract Military sexual trauma (MST) is a serious and pervasive problem among military populations, affecting approximately 16% of military personnel and veterans [1]. Substance use disproportionately affects individuals with a history of MST. Individuals with (vs. without) a history of MST are twice as likely to misuse substances [2-4]. Substance use among military samples has been linked to higher rates of negative consequences across several domains (e.g., health, occupational, legal [5, 6]), including death (e.g., overdose [7], traffic accidents [8], suicide [7, 8]). Further, while understudied among individuals with a history of MST in particular, negative substance use outcomes have been shown to be more severe among trauma-exposed populations, including more severe clinical presentations and poorer treatment prognosis [9, 10]. These findings emphasize the importance of clarifying the association between MST and substance use among military populations. Despite the clinical relevance and public health significance of substance use among military populations, research in this area has relied almost exclusively on cross-sectional designs. Moreover, the vast majority of studies in this area have utilized traditional statistical methods, which are limited in scope and capabilities. These limitations have important clinical implications, as they restrict our ability to specify the exact nature and directionality of the relationship between MST and substance use, thereby affecting how findings are translated into prevention and intervention efforts. The proposed research aims to fill these critical gaps by utilizing the Army STARRS pre/post-deployment study, a large, prospective military dataset to: (1) explicate the directional relation between MST and substance use using a longitudinal dataset, and (2) employ machine learning methods to develop an algorithm to optimize detection of substance use in military personnel with a history of MST. These findings will assist in elucidating the etiology of substance use among this high-risk group, as well as provide a prediction model for clinical use to better target at-risk individuals in this population. This research project will take place within the Department of Psychology at the University of Rhode Island; an institution with a strong history and commitment to health behavior research and methodology. The applicant will have access to sponsors and consultants with expertise in MST, substance use, advanced methodology, and statistical analysis that will facilitate her career objectives to develop increased knowledge and proficiency in (a) sexual trauma (e.g., MST) and substance use in military veterans; (b) grant/manuscript development; (c) statistical and methodological capabilities (i.e., machine learning); and (d) big data. The proposed project uses a timely and innovative approach to advance science on the relation between MST and substance use in military personnel. Addressing substance use in this population is necessary to improve the health of our nation's veterans, and aligns with the mission of the National Institute on Drug Abuse.

项目摘要/摘要 军事性创伤（MST）是军事人口中的一个严重而普遍的问题，影响 大约16％的军事人员和退伍军人[1]。药物使用不成比例地影响个人 拥有MST的历史。 （vs.没有）MST史的个人滥用物质的可能性是 [2-4]。军事样品中的药物使用与整个负面后果的较高率有关 几个领域（例如，健康，职业，法律[5，6]），包括死亡（例如，用药过量[7]，交通事故 [8]，自杀[7，8]）。此外，虽然在有MST史的个体中被研究不足，但 在暴露于创伤的人群中，已经显示出药物使用结果更为严重，包括 更严重的临床表现和较差的治疗预后[9，10]。这些发现强调了 澄清军事人口中MST与物质使用之间的关联的重要性。 尽管军事人口中物质使用的临床相关性和公共卫生意义，但 该领域的研究几乎完全依赖于横截面设计。而且，绝大多数 该领域的研究利用了传统的统计方法，这些方法的范围和能力受到限制。 这些限制具有重要的临床意义，因为它们限制了我们指定确切性质和 MST和物质使用之间关系的方向性，从而影响发现的方式如何翻译 进行预防和干预工作。拟议的研究旨在通过利用 陆军史塔尔斯（Sarters）前/部署后研究，一个大型的，前瞻性的军事数据集：（1）解释方向性 使用纵向数据集和（2）使用机器学习，MST与物质使用之间的关系 开发算法以优化具有历史史的军事人员中使用物质使用的方法 MST。这些发现将有助于阐明这一高风险群体中物质使用的病因 作为为临床使用提供预测模型，以更好地靶向该人群中的高危个体。 该研究项目将在罗德岛大学的心理学系内举行；一个 具有悠久历史和对健康行为研究和方法论的承诺的机构。申请人 将可以与具有MST，物质使用，高级方法论专业知识的赞助商和顾问访问 和统计分析，这将促进她的职业目标，以发展知识和熟练程度 在（a）性创伤（例如MST）和退伍军人中使用物质； （b）赠款/手稿开发； （C） 统计和方法论能力（即机器学习）； （d）大数据。 拟议的项目采用及时，创新的方法来推进MST之间关系的科学 和军事人员使用物质。解决该人群中的物质使用是为了改善的 我们国家退伍军人的健康状况与美国国家药物滥用研究所的任务保持一致。