Ensemble learning of inverse probability weights for marginal structural modeling in large observational datasets.

Ensemble learning of inverse probability weights for marginal structural modeling in large observational datasets.
复制标题

大型观察数据集中边缘结构建模的反概率权重的集合学习。

DOI:
10.1002/sim.6322
复制
发表时间:
2015-01-15
影响因子:
2
通讯作者:
Hernan, Miguel A.
Hernan, Miguel A.
中科院分区:
医学3区
文献类型:
--
作者:
Gruber, Susan;Logan, Roger W.;Jarrin, Inmaculada;Monge, Susana;Hernan, Miguel A.

文献摘要

参考文献

被引文献

相似文献

用于拟合边际结构模型的逆概率权重通常使用逻辑回归进行估计。然而,数据自适应过程可能能够更好地利用测量协变量中可用的信息。通过结合多种算法的预测,集成学习提供了逻辑回归建模的替代方案,以进一步减少估计边际结构模型参数的偏差。我们描述了两种集成学习方法在估计稳定权重方面的应用:超级学习(SL),一种依赖于 V 折交叉验证的集成机器学习方法,以及将数据创建为训练集和验证集的单个分区的集成学习器(EL)。对西班牙两项多中心队列研究(CoRIS 和 CoRIS-MD)的纵向数据进行分析,以估计 HIV 阳性受试者中开始与不开始联合抗逆转录病毒治疗的死亡率风险比。与逻辑回归模型相比,两种集成方法产生的风险比估计值更远离零值,并且置信区间更窄。 EL 的计算时间不到 SL 的一半。我们得出的结论是,在拟合边际结构模型时,使用不同候选算法库的集成学习提供了逆概率权重参数建模的替代方案。对于大型数据集,EL 可以在比 SL 更短的时间内对解决方案空间进行丰富的搜索,并获得可比的结果。
Inverse probability weights used to fit marginal structural models are typically estimated using logistic regression. However a data-adaptive procedure may be able to better exploit information available in measured covariates. By combining predictions from multiple algorithms, ensemble learning offers an alternative to logistic regression modeling to further reduce bias in estimated marginal structural model parameters. We describe the application of two ensemble learning approaches to estimating stabilized weights: super learning (SL), an ensemble machine learning approach that relies on V -fold cross validation, and an ensemble learner (EL) that creates a single partition of the data into training and validation sets. Longitudinal data from two multicenter cohort studies in Spain (CoRIS and CoRIS-MD) were analyzed to estimate the mortality hazard ratio for initiation versus no initiation of combined antiretroviral therapy among HIV positive subjects. Both ensemble approaches produced hazard ratio estimates further away from the null, and with tighter confidence intervals, than logistic regression modeling. Computation time for EL was less than half that of SL. We conclude that ensemble learning using a library of diverse candidate algorithms offers an alternative to parametric modeling of inverse probability weights when fitting marginal structural models. With large datasets, EL provides a rich search over the solution space in less time than SL with comparable results.
DOI: 10.1097/qad.0b013e3283324283
发表时间: 2010-01-02
期刊: AIDS (London, England)
影响因子: --
作者:
HIV-CAUSAL Collaboration;Ray M;Logan R;Sterne JA;Hernández-Díaz S;Robins JM;Sabin C;Bansi L;van Sighem A;de Wolf F;Costagliola D;Lanoy E;Bucher HC;von Wyl V;Esteve A;Casbona J;del Amo J;Moreno S;Justice A;Goulet J;Lodi S;Phillips A;Seng R;Meyer L;Pérez-Hoyos S;García de Olalla P;Hernán MA
通讯作者: Hernán MA
DOI: 10.1037/1082-989x.9.4.403
发表时间: 2004-12-01
影响因子: 7
作者:
McCaffrey, DF;Ridgeway, G;Morral, AR
通讯作者: Morral, AR
DOI: 10.1214/09-ss054
发表时间: 2010-01-01
期刊: STATISTICS SURVEYS
影响因子: 3.3
作者:
Arlot, Sylvain;Celisse, Alain
通讯作者: Celisse, Alain
DOI: 10.1016/j.jclinepi.2013.01.016
发表时间: 2013-08
影响因子: 7.2
作者:
Neugebauer, Romain;Fireman, Bruce;Roy, Jason A.;Raebel, Marsha A.;Nichols, Gregory A.;O'Connor, Patrick J.
通讯作者: O'Connor, Patrick J.
DOI: 10.2202/1557-4679.1181
发表时间: 2010-01-01
影响因子: 1.2
作者:
van der Laan, Mark J.;Gruber, Susan
通讯作者: Gruber, Susan