A model for space-time cluster detection using spatial clusters with flexible temporal risk patterns.

Area of Science:

• Epidemiology
• Biostatistics
• Geographic Information Systems (GIS)

Background:

• Disease mapping is crucial for understanding etiologic insights into potential exposures.
• Existing spatial clustering models need extension to effectively analyze spatio-temporal disease patterns.

Purpose of the Study:

• To extend the Gangnon-Clayton model for spatial clustering to spatio-temporal data.
• To develop a flexible model that accommodates temporal risk patterns, spatial/spatio-temporal heterogeneity, and regional covariates.

Main Methods:

• A Bayesian framework using Markov Chain Monte Carlo (MCMC) for inference.
• Utilizing local Bayes factors with an overly large number of clusters to infer cluster number and location.
• Incorporating circular regions of varying radii to define potential clusters.

Main Results:

• The proposed spatio-temporal model effectively identifies disease clusters and heterogeneity.
• The model demonstrated utility in analyzing female breast cancer mortality data in Japan.
• Simulation studies confirmed the operating characteristics of the developed approach.

Conclusions:

• The extended spatio-temporal model provides a robust method for disease cluster detection.
• This approach enhances etiological investigations by linking disease patterns to specific locations and times.
• The model is adaptable for various disease surveillance and public health research applications.