Extracting and communicating insights about geographical patterns from spatiotemporal disease models

Area of Science:

• Epidemiology
• Biostatistics
• Geographic Information Systems (GIS)

Background:

• Spatiotemporal modeling is crucial for understanding disease dynamics over time and space.
• Existing models often lack detailed uncertainty quantification and actionable metrics.
• This limits the ability to identify granular changes in disease risk.

Purpose of the Study:

• To develop generalizable methods for extracting key metrics from Bayesian relative risk spatiotemporal models.
• To enhance the interpretability of complex spatiotemporal disease patterns.
• To provide actionable insights for public health interventions and prioritization.

Main Methods:

• Described generalizable methods for extracting metrics like relative change magnitude and evidence of non-zero change.
• Introduced analytical approaches for identifying significant cancer rate changes and areas with high rates and large changes.
• Presented a novel Bayesian hierarchical spatiotemporal disease model with flexible implementations for temporal and space-time interaction analysis.

Main Results:

• Demonstrated systematic interpretation of spatiotemporal model outputs.
• Enabled identification of specific cancers or areas with notable rate changes.
• Facilitated visualization of areas with coincident high rates and significant changes.

Conclusions:

• The developed methods enhance the interpretability of spatiotemporal disease models.
• Actionable insights can be generated for public health planning and evaluation.
• Improved understanding of disease risk changes supports evidence-based prioritization and targeted interventions.