Model-based Detection of Spatial Disease Boundaries Using Amortized Bayesian Inference

The article addresses disease boundary analysis, which identifies abrupt changes in health outcomes across geographic boundaries to guide targeted public health interventions. Traditional approaches use Bayesian "wombling" methods and rely on Markov Chain Monte Carlo (MCMC), which creates scalability challenges for large-scale disease surveillance. Researchers applied amortized Bayesian inference (ABI) to accelerate detection of spatial health disparities between neighboring US counties. They introduced a new metric called Residual Disparity Elimination Target, which measures the required reduction in mortality or prevalence to eliminate significant disparities with neighboring regions. The analysis focused on tracheal, bronchus, and lung cancer mortality rates across mainland US counties and achieved results concordant with MCMC analysis while demonstrating improved scalability.