Multiscale detection of localized anomalous structure in aggregate disease incidence data

Area of Science:

• Epidemiology
• Spatial Statistics
• Biostatistics

Background:

• Detecting localized disease outbreaks requires methods sensitive to spatial patterns at various scales.
• Existing spatial disease mapping methodologies need adaptation for robust hypothesis testing.

Purpose of the Study:

• To develop a flexible modeling framework for detecting anomalous spatial disease incidence.
• To enable hypothesis testing for spatially clustered disease variations across multiple scales.

Main Methods:

• Re-casting multiscale disease mapping components for hypothesis testing.
• Linking spatial clustering hypotheses to multiscale parameters within nested spatial partitions.
• Employing a Bayesian hypothesis testing approach with a Poisson measurement model.

Main Results:

• Demonstrated a framework linking spatial disease variations to multiscale parameters.
• Developed a Bayesian methodology for hypothesis testing on these parameters.
• Successfully illustrated the approach on both simulated and real-world disease data.

Conclusions:

• The proposed framework effectively detects anomalous spatial disease patterns at multiple scales.
• The Bayesian hypothesis testing approach provides a robust method for spatial epidemiology.
• This methodology enhances disease surveillance by identifying localized incidence variations.