Spatial correlated incidence modeling with zero inflation.

Area of Science:

• Biostatistics
• Spatial Epidemiology
• Disease Surveillance

Background:

• Disease mapping models are crucial for analyzing spatial disease incidence.
• Zero inflation, characterized by an excess of zero counts, is a common issue in disease incidence datasets, leading to inaccurate estimations.
• This problem often arises from limited survey coverage or inadequate diagnostic tools.

Purpose of the Study:

• To develop and evaluate a zero-inflated disease mapping model to address the issue of excessive zeros in disease incidence data.
• To accurately model spatially correlated disease incidence while accounting for zero inflation.
• To incorporate external covariates and spatial random effects in both zero-inflation and disease incidence processes.

Main Methods:

• A zero-inflated process using Bernoulli indicators was assumed to identify regions with zero inflation.
• A coherent and generative disease mapping model was applied to regions without zero inflation.
• Independent spatial random effects and external covariates were incorporated into both processes. Model estimation was performed using a Markov chain Monte Carlo algorithm.

Main Results:

• Simulation experiments demonstrated the model's effectiveness in handling zero-inflated data.
• The proposed model provided more accurate estimates of disease incidence compared to standard models.
• Application to a Lyme disease dataset from Virginia illustrated the model's practical utility.

Conclusions:

• The developed zero-inflated disease mapping model effectively addresses the challenge of excessive zeros in disease incidence data.
• The model enhances the accuracy of spatial disease mapping and provides a robust framework for epidemiological analysis.
• This approach offers a valuable tool for public health surveillance and understanding disease distribution patterns.