Nonparametric evaluation of dynamic disease risk: a spatio-temporal kernel approach

Area of Science:

• Epidemiology
• Geospatial analysis
• Statistical modeling

Background:

• Understanding disease risk requires joint analysis of spatial and temporal distributions.
• Traditional methods often focus on average spatial patterns, neglecting disease risk trajectories over time.
• Existing approaches mask crucial temporal dynamics in disease spread.

Purpose of the Study:

• To introduce a novel spatio-temporal kernel density estimation (stKDE) method.
• To evaluate disease risk distributions by jointly considering space and time.
• To improve the understanding of epidemiologic trajectories.

Main Methods:

• Employs hybrid kernel (weight) functions for spatio-temporal risk evaluation.
• Leverages information from neighboring points in both space and time.
• Utilizes Monte Carlo simulations for performance assessment.

Main Results:

• The proposed stKDE method significantly outperforms traditional kernel density estimation (trKDE).
• Illustrative examples confirm the superior performance of stKDE over trKDE.
• The method effectively utilizes sample data and borrows information from neighbors.

Conclusions:

• stKDE offers a more comprehensive approach to quantifying disease risk distributions.
• The method enhances the understanding of spatio-temporal disease patterns and epidemiologic trajectories.
• There are opportunities for future improvements and extensions of the stKDE method.