A method for comparing multiple imputation techniques: A case study on the U.S. national COVID cohort collaborative.
A method for comparing multiple imputation techniques: A case study on the U.S. national COVID cohort collaborative.
复制标题
DOI：
10.1016/j.jbi.2023.104295
复制
发表时间：
2023-03
影响因子：
4.5
通讯作者：
Wilkins, Kenneth J.
中科院分区：
文献类型：
--
作者：
Casiraghi, Elena;Wong, Rachel;Hall, Margaret;Coleman, Ben;Notaro, Marco;Evans, Michael D.;Tronieri, Jena S.;Blau, Hannah;Laraway, Bryan;Callahan, Tiffany J.;Chan, Lauren E.;Bramante, Carolyn T.;Buse, John B.;Moffitt, Richard A.;Sturmer, Til;Johnson, Steven G.;Shao, Yu Raymond;Reese, Justin;Robinson, Peter N.;Paccanaro, Alberto;Valentini, Giorgio;Huling, Jared D.;Wilkins, Kenneth J.
关键词：
事实证明，从电子健康记录获得的医疗保健数据集对于评估患者预测因子与感兴趣的结果之间的关联非常有用。然而，这些数据集往往在很大比例的情况下存在缺失值，其删除可能会引入严重的偏差。为了在假定的缺失机制下恢复缺失信息，已经提出了几种多重插值算法。每种算法都有各自的优点和缺点，目前对于在给定情况下哪种多重插值算法效果最好还没有达成共识。此外，每种算法的参数选择和与数据相关的建模选择也是至关重要和具有挑战性的。在本文中，我们提出了一个新的框架，在统计分析的背景下，对处理缺失数据的策略进行数值评估，特别关注多重imputation技术。我们在国家COVID队列协作（N3C） Enclave提供的大型2型糖尿病患者队列中证明了该方法的可行性，在该队列中，我们探索了各种患者特征对COVID-19相关结局的影响。我们的分析包括经典的多重插值技术以及简单的全情况逆概率加权模型。大量的实验表明，我们的方法可以有效地突出我们的案例研究中最有前途和性能的丢失数据处理策略。此外，我们的方法可以更好地理解不同模型的行为，以及当我们修改它们的参数时它是如何变化的。我们的方法是通用的，可以应用于不同的研究领域和包含异构类型的数据集。
Healthcare datasets obtained from Electronic Health Records have proven to be extremely useful for assessing associations between patients’ predictors and outcomes of interest. However, these datasets often suffer from missing values in a high proportion of cases, whose removal may introduce severe bias. Several multiple imputation algorithms have been proposed to attempt to recover the missing information under an assumed missingness mechanism. Each algorithm presents strengths and weaknesses, and there is currently no consensus on which multiple imputation algorithm works best in a given scenario. Furthermore, the selection of each algorithm’s parameters and data-related modeling choices are also both crucial and challenging. In this paper we propose a novel framework to numerically evaluate strategies for handling missing data in the context of statistical analysis, with a particular focus on multiple imputation techniques. We demonstrate the feasibility of our approach on a large cohort of type-2 diabetes patients provided by the National COVID Cohort Collaborative (N3C) Enclave, where we explored the influence of various patient characteristics on outcomes related to COVID-19. Our analysis included classic multiple imputation techniques as well as simple complete-case Inverse Probability Weighted models. Extensive experiments show that our approach can effectively highlight the most promising and performant missing-data handling strategy for our case study. Moreover, our methodology allowed a better understanding of the behavior of the different models and of how it changed as we modified their parameters. Our method is general and can be applied to different research fields and on datasets containing heterogeneous types.
登录
查看更多内容
影响因子：
5
作者：
Burgette, Lane F.;Reiter, Jerome P.
通讯作者：
Reiter, Jerome P.
影响因子：
11.1
作者：
Deer RR;Rock MA;Vasilevsky N;Carmody L;Rando H;Anzalone AJ;Basson MD;Bennett TD;Bergquist T;Boudreau EA;Bramante CT;Byrd JB;Callahan TJ;Chan LE;Chu H;Chute CG;Coleman BD;Davis HE;Gagnier J;Greene CS;Hillegass WB;Kavuluru R;Kimble WD;Koraishy FM;Köhler S;Liang C;Liu F;Liu H;Madhira V;Madlock-Brown CR;Matentzoglu N;Mazzotti DR;McMurry JA;McNair DS;Moffitt RA;Monteith TS;Parker AM;Perry MA;Pfaff E;Reese JT;Saltz J;Schuff RA;Solomonides AE;Solway J;Spratt H;Stein GS;Sule AA;Topaloglu U;Vavougios GD;Wang L;Haendel MA;Robinson PN
通讯作者：
Robinson PN
DOI：
10.1109/access.2020.3034032
发表时间：
2020
期刊：
IEEE access : practical innovations, open solutions
影响因子：
--
作者：
Casiraghi E;Malchiodi D;Trucco G;Frasca M;Cappelletti L;Fontana T;Esposito AA;Avola E;Jachetti A;Reese J;Rizzi A;Robinson PN;Valentini G
通讯作者：
Valentini G
DOI：
10.1093/jamia/ocaa196
发表时间：
2021-03-01
影响因子：
6.4
作者：
Haendel, Melissa A.;Chute, Christopher G.;Gersing, Ken R.
通讯作者：
Gersing, Ken R.
影响因子：
12.7
作者：
Bramante CT;Buse J;Tamaritz L;Palacio A;Cohen K;Vojta D;Liebovitz D;Mitchell N;Nicklas J;Lingvay I;Clark JM;Aronne LJ;Anderson E;Usher M;Demmer R;Melton GB;Ingraham N;Tignanelli CJ
通讯作者：
Tignanelli CJ