Cluster pattern detection in spatial data based on Monte Carlo inference.

Area of Science:

• Spatial statistics
• Point process modeling
• Epidemiology

Background:

• Cluster detection is crucial for understanding spatial patterns in various fields.
• Existing methods may struggle with data heterogeneity and complex spatial interactions.
• Identifying spatially specified regions formed by overlapping random disks is a key challenge.

Purpose of the Study:

• To propose a new statistical model for detecting spatial clusters.
• To model cluster regions as overlapping random disks driven by a marked point process.
• To apply the method to animal epidemiology data and address practical challenges.

Main Methods:

• Utilizing a marked point process with two components: inhomogeneous Poisson process for disk locations and superposition of area-interaction and pairwise interaction processes for disk interactions.
• Developing statistical descriptors and methods for assessing the probability and presence of clusters.
• Applying the model to real-world spatial data from animal epidemiology.

Main Results:

• The proposed method effectively detects spatially specified regions indicative of clusters.
• The model accounts for data heterogeneity and incorporates smoothing effects.
• It provides statistical descriptors and enables testing for cluster presence.

Conclusions:

• The marked point process model offers a robust framework for spatial cluster detection.
• This approach enhances the analysis of spatial data, particularly in epidemiology.
• The method provides a comprehensive tool for understanding and quantifying spatial clustering.