Hierarchical multivariate mixture generalized linear models for the analysis of spatial data: An application to...

Area of Science:

• Biostatistics
• Epidemiology
• Spatial Analysis

Background:

• Disease mapping traditionally focuses on single diseases.
• Simultaneous modeling of related diseases offers epidemiological and statistical advantages.
• Existing multivariate spatial models often assume uniform population distributions, which can be limiting.

Purpose of the Study:

• To propose a novel hierarchical multivariate mixture generalized linear model.
• To simultaneously analyze spatial Normal and non-Normal outcomes.
• To address the limitation of assuming a single distribution across diverse population densities.

Main Methods:

• Developed a hierarchical multivariate mixture generalized linear model.
• Applied the model to analyze spatial data with both Normal and non-Normal outcomes.
• Utilized esophageal and lung cancer death data from Minnesota for application.

Main Results:

• The proposed model demonstrated superior performance compared to models assuming a single distribution.
• The application to Minnesota cancer data highlighted the benefits of allowing different distributions for different counties.
• A simulation study further validated the effectiveness of the developed approach.

Conclusions:

• The hierarchical multivariate mixture generalized linear model effectively handles spatial and multivariate dependencies.
• Allowing for varying population distributions improves the accuracy of disease mapping for related outcomes.
• This approach offers a more flexible and robust tool for public health surveillance and statistical analysis.