Technology Category
Reasoning under Uncertainty: Relational Probabilistic ModelsMachine Learning: Calibration & Uncertainty QuantificationKnowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
A spatial random forest algorithm for population-level epidemiological risk assessment
🤖 AI Summary
This study addresses the limitations of traditional spatial epidemiological models, which rely on prespecified functional forms and interaction structures for confounders and thus struggle to capture nonlinearities and complex interactions in data. The authors propose SPAR-Forest-ERF, a novel algorithm that integrates random forests into spatial epidemiological inference for the first time, combining them with a Bayesian spatial autocorrelation model. This approach automatically learns nonlinear and high-order interactions among confounders while preserving an interpretable exposure–response function. A full uncertainty quantification is achieved through a principled uncertainty propagation mechanism, and inference stability is ensured via an adaptive stopping criterion. The method supports flexible exposure–response specifications and, when applied to the 2022 Scottish census data, effectively quantifies the impact of air pollution on self-rated health, yielding interpretable results with comprehensive uncertainty estimates.
Technology Category
Application Category
📝 Abstract
Spatial epidemiology identifies the drivers of elevated population-level disease risks, using disease counts, exposures and known confounders at the areal unit level. Poisson regression models are typically used for inference, which incorporate a linear/additive regression component and allow for unmeasured confounding via a set of spatially autocorrelated random effects. This approach requires the confounder interactions and their functional relationships with disease risk to be specified in advance, rather than being learned from the data. Therefore, this paper proposes the SPAR-Forest-ERF algorithm, which is the first fusion of random forests for capturing non-linear and interacting confounder-response effects with Bayesian spatial autocorrelation models that can estimate interpretable exposure response functions (ERF) with full uncertainty quantification. Methodologically, we extend existing methods set in a prediction context by propagating uncertainty between both the ML and statistical models, developing a new stopping criteria designed to ensure the stability of the primary inferential target, and incorporating a range of different ERFs for maximum model flexibility. This methodology is motivated by a new study quantifying the impact of air pollution concentrations on self-rated health in Scotland, using data from the recently released 2022 national census.
Problem
Research questions and friction points this paper is trying to address.
spatial epidemiology
confounder interactions
non-linear effects
disease risk assessment
exposure response functions
Innovation
Methods, ideas, or system contributions that make the work stand out.
SPAR-Forest-ERF
spatial random forest
exposure-response function
Bayesian spatial autocorrelation
uncertainty quantification
🔎 Similar Papers
No similar papers found.
