| 向韧,戴文杰,熊元,吴鑫,杨艳芳,王玲,戴志辉,李娇,刘爱忠.有向无环图在因果推断控制混杂因素中的应用[J].Chinese journal of Epidemiology,2016,37(7):1035-1038 |
| 有向无环图在因果推断控制混杂因素中的应用 |
| Application of directed acyclic graphs in control of confounding |
| Received:November 18, 2015 |
| DOI：10.3760/cma.j.issn.0254-6450.2016.07.025 |
| KeyWord: 病因学研究 有向无环图 混杂因素 因果关系 |
| English Key Word: Etiology study Directed acyclic graph Confounder Causality |
| FundProject: |
| Author Name | Affiliation | E-mail | | Xiang Ren | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Dai Wenjie | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Xiong Yuan | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Wu Xin | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Yang Yanfang | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Wang Ling | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Dai Zhihui | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Li Jiao | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | | | Liu Aizhong | Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha 410008, China | lazroy@live.cn |
|
| Hits: 8787 |
| Download times: 4943 |
| Abstract: |
| 观察性研究是流行病学病因研究中最常用的方法之一，但在因果推断时混杂因素往往会歪曲暴露与结局的真实因果关联。为了消除混杂，选择变量调整是关键所在。有向无环图能够将复杂的因果关系可视化，提供识别混杂的直观方法，将识别混杂转变成识别最小充分调整集。一方面有向无环图可以选择调整更少的变量，增加分析的统计效率；另一方面有向无环图识别的最小充分调整集可以避开未被测量或有缺失值的变量。总之，有向无环图有助于充分揭示真实的因果关系。 |
| English Abstract: |
| Observational study is a method most commonly used in the etiology study of epidemiology, but confounders, always distort the true causality between exposure and outcome when local inferencing. In order to eliminate these confounding, the determining of variables which need to be adjusted become a key issue. Directed acyclic graph (DAG) could visualize complex causality, provide a simple and intuitive way to identify the confounding, and convert it into the finding of the minimal sufficient adjustment for the control of confounding. On the one hand, directed acyclic graph can choose less variables, which increase statistical efficiency of the analysis. On the other hand, it could help avoiding variables that is not measured or with missing values. In a word, the directed acyclic graph could facilitate the reveal of the real causality effectively. |
|
View Fulltext
Html FullText
View/Add Comment Download reader |
| Close |
|
|
|