Holland R C, Jones G, Benschop J
The Institute of Fundamental Sciences, Massey University,Palmerston North,New Zealand.
Molecular Epidemiology and Public Health Laboratory, Institute of Veterinary, Animal and Biomedical Sciences, Massey University,Palmerston North,New Zealand.
Epidemiol Infect. 2015 Jun;143(8):1777-88. doi: 10.1017/S0950268814002854. Epub 2014 Oct 23.
The search for an association between disease incidence and possible risk factors using surveillance data needs to account for possible spatial and temporal correlations in underlying risk. This can be especially difficult if there are missing values for some important covariates. We present a case study to show how this problem can be overcome in a Bayesian analysis framework by adding to the usual spatio-temporal model a component for modelling the missing data.
利用监测数据探寻疾病发病率与潜在风险因素之间的关联时，需要考虑潜在风险中可能存在的空间和时间相关性。如果某些重要协变量存在缺失值，这可能会尤其困难。我们通过一个案例研究展示了在贝叶斯分析框架下，如何通过在常规时空模型中添加一个用于对缺失数据建模的组件来克服这一问题。
