An Algorithm for Creating Virtual Controls Using Integrated and Harmonized Longitudinal Data.
An Algorithm for Creating Virtual Controls Using Integrated and Harmonized Longitudinal Data.
复制标题
DOI：
10.1177/0163278718772882
复制
发表时间：
2018-06
影响因子：
2.9
通讯作者：
Ip EH
中科院分区：
文献类型：
--
作者：
Hansen WB;Chen SH;Saldana S;Ip EH
我们介绍了一种策略，用于创建虚拟控制组-通过计算机算法生成的情况下，聚合时，可以作为实验比较器，现场控制难以招募，如当程序被广泛传播和随机化是不可行的。我们整合和协调了1980年至2010年数十年来8个存档的纵向数据集的数据。总的来说，这些研究检查了许多心理社会变量，并评估了过去30天的酒精，香烟和大麻使用情况。使用来自两项存档随机对照试验的额外治疗组和对照组数据来测试虚拟对照算法。两项随机对照试验都评估了意图、规范信念和价值观以及过去30天的酒精、香烟和大麻使用情况。我们开发了一种算法，该算法使用来自综合数据集的百分位数得分来创建特定年龄和性别的潜在心理社会得分。该算法匹配治疗病例在预测试时观察到的心理社会分数，以创建一个虚拟对照病例，该虚拟对照病例基于年龄相关的变化象征性地“成熟”，保持虚拟病例的百分位数不变。虚拟对照匹配治疗病例发生率，消除差异磨损作为有效性的威胁。虚拟情况下物质使用估计从虚拟情况下的潜在的社会心理评分使用逻辑回归系数来自分析治疗组。对虚拟病例取平均值可得出患病率的群体估计值。建立了两个标准来评估虚拟对照病例的充分性：（1）虚拟对照组试验前药物流行率应与治疗组相匹配，（2）虚拟对照组药物流行模式随时间的变化应与现场对照相匹配。该算法成功匹配了两项RCT的预测患病率。观察到患病率增加;尽管实时和虚拟控制结果之间存在差异。这项研究提供了一个初步的框架，使用一个逐步的程序，现在可以使用其他预防试验数据进行修改和验证创建虚拟控制。
We introduce a strategy for creating virtual control groups – cases generated through computer algorithms that, when aggregated, may serve as experimental comparators where live controls are difficult to recruit, such as when programs are widely disseminated and randomization is not feasible. We integrated and harmonized data from eight archived longitudinal adolescent-focused datasets spanning the decades from 1980 to 2010. Collectively, these studies examined numerous psychosocial variables and assessed past 30-day alcohol, cigarette and marijuana use. Additional treatment and control group data from two archived randomized control trials were used to test the virtual control algorithm. Both RCTs assessed intentions, normative beliefs and values as well as past 30-day alcohol, cigarette and marijuana use. We developed an algorithm that used percentile scores from the integrated dataset to create age- and gender-specific latent psychosocial scores. The algorithm matched treatment case observed psychosocial scores at pretest to create a virtual control case that figuratively “matured” based on age-related changes, holding the virtual case’s percentile constant. Virtual controls matched treatment case occurrence, eliminating differential attrition as a threat to validity. Virtual case substance use was estimated from the virtual case’s latent psychosocial score using logistic regression coefficients derived from analyzing the treatment group. Averaging across virtual cases created group estimates of prevalence. Two criteria were established to evaluate the adequacy of virtual control cases: (1) virtual control group pretest drug prevalence rates should match those of the treatment group, (2) virtual control group patterns of drug prevalence over time should match live controls. The algorithm successfully matched pretest prevalence for both RCTs. Increases in prevalence were observed; although, there were discrepancies between live and virtual control outcomes. This study provides an initial framework for creating virtual controls using a step-by-step procedure that can now be revised and validated using other prevention trial data.
登录
查看更多内容
影响因子：
7
作者：
Curran, Patrick J.;Hussong, Andrea M.
通讯作者：
Hussong, Andrea M.
影响因子：
1.6
作者：
Campbell, DT
通讯作者：
Campbell, DT
影响因子：
2.3
作者：
Doiron D;Burton P;Marcon Y;Gaye A;Wolffenbuttel BHR;Perola M;Stolk RP;Foco L;Minelli C;Waldenberger M;Holle R;Kvaløy K;Hillege HL;Tassé AM;Ferretti V;Fortier I
通讯作者：
Fortier I
影响因子：
3.7
作者：
Cooper R;Hardy R;Aihie Sayer A;Ben-Shlomo Y;Birnie K;Cooper C;Craig L;Deary IJ;Demakakos P;Gallacher J;McNeill G;Martin RM;Starr JM;Steptoe A;Kuh D;HALCyon study team
通讯作者：
HALCyon study team
影响因子：
8.8
作者：
Booth, C. M.;Tannock, I. F.
通讯作者：
Tannock, I. F.