贝叶斯因果关系网络模型在断面调查数据中的应用

Application of Bayesian network in cross-sectional data

目的验证贝叶斯网络能否从横断面调查数据的分析中得到因果关系的信息。方法通过从国外数据库获取的调查数据，利用R语言进行贝叶斯网络因果关系的结构分析。结果在无先验信息情况下，从网络结构中得出，胰岛素对空腹血糖、收缩压有一定的调控关系，HDL对总胆固醇的代谢也起到一定的调控作用，以上关系与文献研究结果一致；总胆固醇、LDL、HDL、apoB均指向甘油三酯，此关系有待解释。结论与其他统计方法相比，贝叶斯网络能够从横断面调查数据中得到因果关系的重要信息，为了解疾病机理、探索疾病病因提供了新的有效方法。

Objective To testify that Bayesian network can be used to analyze the cross-sectional data. Methods Obtaining investigation data from the overseas database, then analyzing the causation structure with R. Results Without prior knowledge, Bayesian network showed that insulin regulated blood fasting glucose and systolic pressure. Meanwhile, that HDL regulated TC was also shown, which consisted with the existing literature. It was to be explained why total cholesterol, LDL, HDL, apoB all pointed to triglyceride. Conclusion Comparing with other methods, Bayesian network can learn cause-effect relationship from investigative data, which may help us understand disease mechanism and explore the disease causes.