An Integrative Approach to Control Group Creation for Prevention Research

项目摘要 该第二阶段SBIR项目的目标是开发一种用于创建算法生成的控件的方法 组-虚拟控件。这些将由具有两个特征的合成案例组成：(1)前测 与任何可能出现的预防治疗组相似，以及(2)随时间的变化模式 与酒精和药物使用发展的正常过程非常相似。酒精和毒品预防领域 是一个已经显著成熟并拥有丰富的数据资源，可以为此目的而使用的公司。值得注意的是， 许多学校和社区的疗效研究都在随机对照试验中使用了对照组。 其他病因学和流行病学研究也收集了类似的数据。考虑到大量的数据 可供分析，现在可以对青少年饮酒和吸毒的开始进行建模。我们 建议从以前的研究中收集数据，并开发可用于预测 开始酗酒和吸毒。将开发一种算法，该算法将创建综合控制案例以匹配 治疗组的人口学数据，并预先测试中介变量分数，然后估计未来的药物使用情况。 这种方法的几个好处是可以预期的。酒精和毒品预防研究人员和从业者将 能够使用这种方法来测试传播干预措施的有效性，并快速评估 新的和替代的预防干预措施的潜力。具体地说，此方法将提供一种方法来 评估随机化时传播的酒精和毒品预防计划的有效性 不可能发生。这种方法还将使评估新的和适应的计划更容易，因为减少了 招聘、受试者留存和现场数据收集方面的挑战，以及经常出现的不对等预测 发生在L教室、学校和社区分配条件的时候。在这个项目中，我们将(1)收集 来自以前针对青少年的纵向药物和酒精研究项目的数据；(2)协调 这些数据与预定义的编码方案相匹配，并将它们包括在综合参考数据库中； (3)改进和验证在第一阶段开发的统计模型，以确保算法用于生成 虚拟控制案例是有效的和可复制的；(4)准备一个基于Web的系统，允许虚拟控制 集团技术将最大限度地自动化；以及(5)对传播干预措施进行从头开始的现场试验 以证明该系统在提供有意义的比较方面的可行性。

PROJECT ABSTRACT The goal of this Phase II SBIR project is to develop a method for creating algorithmically generated control groups – virtual controls. These will be composed of synthetic cases that have two characteristics: (1) pretest similarity to whatever prevention treatment group may present itself and (2) patterns of change over time that closely mimic the normal course of alcohol and drug use development. The field of alcohol and drug prevention is one that has significantly matured and has rich data resources that can be employed to this end. Notably, numerous school and community efficacy studies have employed control groups in randomized control trials. Additional etiological and epidemiological studies have collected similar data. Given the large amount of data available for analysis, it is now possible to model the onset of alcohol and drug use among adolescents. We propose to gather data from previous studies and develop statistical models that can be used to predict the onset of alcohol and drug use. An algorithm will be developed that will create integrative control cases to match to a treatment group’s demographics and pretest mediating variable scores and then estimate future drug use. Several benefits of this method are anticipated. Alcohol and drug prevention researchers and practitioners will be able to use this approach to test the effectiveness of disseminated interventions and to quickly evaluate the potential of new and alternative prevention interventions. Specifically, this method will provide a means to evaluate the effectiveness of alcohol and drug prevention programs that are disseminated when randomization cannot occur. This method will also make evaluating new and adapted programs easier by reducing the challenges of recruitment, subject retention, and onsite data collection and pretest non-equivalence that often occur when l classrooms, schools and communities are assigned to condition. In this project, we will (1) gather data from previous longitudinal drug and alcohol research projects focused on adolescents; (2) harmonize these data to match a predefined coding scheme and include them in an integrated referential database; (3) refine and validate statistical models developed during Phase I to ensure algorithms used to generate virtual control cases are valid and replicable; (4) prepare a web-based system that will allow virtual control group technology to be maximally automated; and (5) conduct de novo field trials of disseminated interventions to demonstrate the viability of the system for providing meaningful comparisons.