Reducing racial disparities in the treatment of opioid use disorder using machine learning-based causal analysis

使用基于机器学习的因果分析减少阿片类药物使用障碍治疗中的种族差异

基本信息

批准号：
10557201
负责人：
Mathew Vinhhoa Kiang
金额：
$ 24.9万
依托单位：
STANFORD UNIVERSITY
依托单位国家：
美国
项目类别：
财政年份：
2020
资助国家：
美国
起止时间：
2020-07-01 至 2025-01-31
项目状态：
未结题

来源：
https://reporter.nih.gov/project-details/10557201
关键词：
Address Affect Affordable Care Act Age American Area Big Data Biometry Black American Black Populations Black race Buprenorphine Caring Characteristics Client satisfaction Clinical Treatment Communities Community Surveys Complement Computerized Medical Record Critical Care Data Databases Development Plans Effectiveness Eligibility Determination Epidemiology Equity Ethnic Origin Ethnic Population Face Foundations Health Health Personnel Health Resources Health Services Accessibility Healthcare Systems Individual Intervention Machine Learning Measures Medical Mental Health Mentorship Methods Minority Groups Modeling Natural experiment Opioid Patients Perception Persons Pharmaceutical Preparations Phase Population Prediction of Response to Therapy Provider Public Health Race Reduce health disparities Research Specialist Substance Use Disorder Therapeutic Training Treatment outcome Veterans Veterans Health Administration Visit Work barrier to care black patient career career development caucasian American data warehouse design electronic data ethnic disparity experience forest improved improved outcome innovation machine learning method medication for opioid use disorder mortality multidisciplinary multiple data sources novel strategies opioid overdose opioid use disorder racial bias racial disparity racial population skills social standard of care structural determinants success treatment disparity

项目摘要

PROJECT SUMMARY The opioid overdose crisis emerged in predominantly White communities, but the opioid-related mortality rate is increasing most rapidly in the Black population. A key driver of the crisis is opioid use disorder, which affects over 2 million Americans. Despite their effectiveness, medications for opioid use disorder remain underused, especially among Black Americans. Compared to White Americans, Black Americans have lower access to medications for opioid use disorder, are one-third as likely to initiate treatment, and have lower retention in care. Black Americans face unique structural obstacles to care, such as mistrust of the health care system, lack of representation among medical providers, and racially-biased providers’ perceptions. There is a critical gap in our understanding of the structural factors associated with treatment initiation and retention in care for Black patients with OUD. The scientific objective of this research plan is to identify modifiable structural factors at the community, provider, and facility levels that affect treatment initiation and retention in care for opioid use disorder in the Black population. This innovative project proposes to leverage machine learning-based causal inference methods with a combination of large national electronic medical records, corporate data warehouses, and publicly available data. By combining multiple data sources, this project will empirically evaluate modifiable factors such as provider characteristics (e.g., years of experience, patient satisfaction scores), facility characteristics (e.g., mental health staffing to patient ratios, number of buprenorphine-eligible prescribers), and patient-provider characteristics (e.g., number of previous visits or interactions). While focused on promoting equitable access to treatment for opioid use disorder in Black Americans, the public health implications of this proposal are expected to apply broadly to ameliorate the overall health burden of substance use disorders and reduce health disparities. This research plan is complemented by a career development plan that builds on the applicant’s background in epidemiology and biostatistics. Specifically, this career development plan outlines new training in three areas: (1) the clinical treatment of opioid use disorder, (2) analysis of the massive data of electronic medical records, and (3) machine learning-based causal inference methods. The combined research and training plan will prepare the applicant for a successful independent research career identifying, evaluating, and implementing multilevel interventions to reduce racial/ethnic inequalities in treatment for substance use disorders.

项目摘要阿片类药物过量危机主要是白人社区，但与阿片类药物有关黑人人口中的死亡率最快。危机的主要驱动力是阿片类药物使用障碍，影响超过200万美国人。尽管它们有效，但阿片类药物使用障碍的药物仍然未充分利用，尤其是在黑人美国人中。与白人美国人相比，黑人美国人的较低获得阿片类药物使用障碍的药物，可能启动治疗的可能性三分之一，并且较低保留护理。黑人美国人面临着独特的结构性护理障碍，例如对医疗保健的不信任系统，医疗提供者缺乏代表性以及大致偏见的提供者的看法。有一个我们对与治疗计划相关的结构因素的理解和保留率的关键差距照顾黑人患者。该研究计划的科学目标是确定可修改的社区，提供商和设施水平的结构性因素，影响治疗计划和保留护理黑人人群中阿片类药物使用障碍。这项创新的项目提案要利用机器基于学习的因果推理方法，结合了大型国家电子病历，公司数据仓库和公开数据。通过组合多个数据源，该项目将经验评估可修改因素，例如提供商特征（例如，经验年，患者满意度得分），设施特征（例如，心理健康人员与患者比例，数量丁丙诺啡符合条件的处方者）和患者提供的特征（例如，先前的访问次数或互动）。虽然专注于促进黑色阿片类药物使用障碍的公平治疗美国人，这一提案的公共卫生影响预计将广泛适用于改善物质使用障碍的总体健康燃烧并减少健康差异。该研究计划是由一项职业发展计划完成，该计划以申请人的流行病学背景为基础生物统计学。特别是，该职业发展计划在三个领域概述了新培训：（1）临床治疗阿片类药物使用障碍，（2）分析电子病历的大量数据，以及（3）基于机器学习的因果推理方法。联合研究和培训计划将为成功的独立研究职业的申请人识别，评估和实施多级减少种族/族裔不平等的干预措施，以治疗药物使用障碍。