A comparison of disease mapping of dengue incidences in Tamil Nadu using Bayesian conditional autoregressive model and intrinsic conditional autoregressive model

Abstract

Spatial regression models allow a more accurate description of true disease rates by borrowing information from neighboring regions which would help to mitigate the effects of sparsely populated regions and provide better inference. Perhaps the most conspicuous manner of modeling spatial dependence is to introduce spatially associated random effects within a Bayesian hierarchical setting. The Bayesian modeling and inferential framework are flexible and extremely rich in its capabilities to accumulate various scientific hypotheses and assumptions. Conditional Autoregressive Models are quite useful in representing spatial autocorrelation in data relating to a set of non-overlapping geographical units, which could be meaningfully applied in several problems in agriculture and epidemiology. Bayesian Conditional Autoregressive model is a disease mapping methodology used for smoothening of the relative risk of disease. The Intrinsic Conditional Autoregressive model is a special case of the standard conditional autoregressive model in which the complete spatial dependencies are assumed. The present study aims to compare the Bayesian spatial conditional autoregressive models (CAR and ICAR) for exploring the spatial patterns of Dengue cases in Tamil Nadu.