Hypotheses testing procedures based on quadratic inference functions are proposed to test whether two Gaussian Bayesian networks are differential in structure, strength of associations between nodes, or both. Bootstrap procedures are developed to estimate $p$-values to quantify the statistical significance of the tests. Operating characteristics of these testing procedures are investigated using synthetic data in simulation experiments. Additionally, the proposed methods are applied to flow cytometry data from a designed experiment, and data of bile acids from an observational study in the Alzheimer’s Disease Neuroimaging Initiative.

中文翻译：

---

基于二次推断函数的差分高斯贝叶斯网络测试

提出了基于二次推断函数的假设检验程序，以检验两个高斯贝叶斯网络在结构，节点之间的关联强度或两者之间是否存在差异。制定引导程序以估算$p$值以量化测试的统计显着性。在模拟实验中使用合成数据研究了这些测试程序的操作特性。此外，将拟议的方法应用于设计实验的流式细胞仪数据，以及阿尔茨海默氏病神经影像学计划的一项观察性研究中的胆汁酸数据。