Comparison of different software implementations for spatial disease mapping

Vranckx, Maren; Neyens, Thomas; Faes, Christel

doi:10.1016/j.sste.2019.100302

Cited by 10 publications

(4 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We performed the Bayesian analysis using an integrated nested Laplace approximation (INLA) program in R software [ 21 ]. The deterministic algorithm approach for Bayesian inference in INLA has been proven to reduce the computing time and provides accurate results [ 22 , 23 ]. In Bayesian inference, prior distributions for parameters to be estimated were specified before modelling was commenced.…”

Section: Discussionmentioning

confidence: 99%

A practical illustration of spatial smoothing methods for disconnected regions with INLA: spatial survey on overweight and obesity in Malaysia

2023

View full text Add to dashboard Cite

Background National prevalence could mask subnational heterogeneity in disease occurrence, and disease mapping is an important tool to illustrate the spatial pattern of disease. However, there is limited information on techniques for the specification of conditional autoregressive models in disease mapping involving disconnected regions. This study explores available techniques for producing district-level prevalence estimates for disconnected regions, using as an example childhood overweight in Malaysia, which consists of the Peninsular and Borneo regions separated by the South China Sea. We used data from Malaysia National Health and Morbidity Survey conducted in 2015. We adopted Bayesian hierarchical modelling using the integrated nested Laplace approximation (INLA) program in R-software to model the spatial distribution of overweight among 6301 children aged 5–17 years across 144 districts located in two disconnected regions. We illustrate different types of spatial models for prevalence mapping across disconnected regions, taking into account the survey design and adjusting for district-level demographic and socioeconomic covariates. Results The spatial model with split random effects and a common intercept has the lowest Deviance and Watanabe Information Criteria. There was evidence of a spatial pattern in the prevalence of childhood overweight across districts. An increasing trend in smoothed prevalence of overweight was observed when moving from the east to the west of the Peninsular and Borneo regions. The proportion of Bumiputera ethnicity in the district had a significant negative association with childhood overweight: the higher the proportion of Bumiputera ethnicity in the district, the lower the prevalence of childhood overweight. Conclusion This study illustrates different available techniques for mapping prevalence across districts in disconnected regions using survey data. These techniques can be utilized to produce reliable subnational estimates for any areas that comprise of disconnected regions. Through the example, we learned that the best-fit model was the one that considered the separate variations of the individual regions. We discovered that the occurrence of childhood overweight in Malaysia followed a spatial pattern with an east–west gradient trend, and we identified districts with high prevalence of overweight. This information could help policy makers in making informed decisions for targeted public health interventions in high-risk areas.

show abstract

Section: Discussionmentioning

confidence: 99%

A practical illustration of spatial smoothing methods for disconnected regions with INLA: spatial survey on overweight and obesity in Malaysia

2023

View full text Add to dashboard Cite

show abstract

“…With hierarchically structured models, including several components with different structures at the latent layer, INLA is typically faster and simpler to tune than simulation-based Markov chain Monte Carlo (MCMC) methods (Illian, Sørbye and Rue (2012), Opitz (2017), Rue and Held (2005), Rue, Martino and Chopin (2009), Rue et al (2016)). INLA is an approximate inference method, but the approximation quality is generally superior to MCMC-based approaches when using similar computation times (Teng, Nathoo and Johnson (2017), Vranckx, Neyens and Faes (2019)), especially in cases where achieving good mixing properties within MCMC is difficult. The INLA method can be applied to models where the observations are conditionally independent with respect to a latent multivariate Gaussian random vector which parameterizes the observation model.…”

Section: Approximate Bayesian Inferencementioning

confidence: 99%

High-resolution Bayesian mapping of landslide hazard with unobserved trigger event

et al. 2022

View full text Add to dashboard Cite

Statistical models for landslide hazard enable mapping of risk factors and landslide occurrence intensity by using geomorphological covariates available at high spatial resolution. However, the spatial distribution of the triggering event (e.g., precipitation or earthquakes) is often not directly observed. In this paper we develop Bayesian spatial hierarchical models for point patterns of landslide occurrences using different types of log-Gaussian Cox processes. Starting from a competitive baseline model that captures the unobserved precipitation trigger through a spatial random effect at slope unit resolution, we explore novel complex model structures that take clusters of events arising at small spatial scales into account as well as nonlinear or spatially-varying covariate effects. For a 2009 event of around 5000 precipitation-triggered landslides in Sicily, Italy, we show how to fit our proposed models efficiently, using the integrated nested Laplace approximation (INLA), and rigorously compare the performance of our models both from a statistical and applied perspective. In this context we argue that model comparison should not be based on a single criterion and that different models of various complexity may provide insights into complementary aspects of the same applied problem. In our application our models are found to have mostly the same spatial predictive performance, implying that key to successful prediction is the inclusion of a slope-unit resolved random effect capturing the precipitation trigger. Interestingly, a parsimonious formulation of space-varying slope effects reflects a physical interpretation of the precipitation trigger: in subareas with weak trigger, the slope steepness is shown to be mostly irrelevant.

show abstract

“…Specific values of α 1 = α 2 = 0.001 for this prior distribution are often employed in many applications (see [17]), so that ψ ∼ G(0.001, 0.001), which, given that its mean is equal to 1 and its variance equal to 1000, a large value, it can be considered as a vague prior. Alternative frequently used values that can be found in the literature are, α 1 = 1 and α 2 = 0.01, in Vranckx, Neyens and Faes [30], α 1 = 0.05, α 2 = 5 × 10 −4 in Best, Richardson and Thomson [18], α 1 = 1 and α 2 = 0.5 in Carroll, Lawson, Faes, Kirby, Aregay and Watjou [29], α 1 = α 2 = 1 × 10 −4 in Cepeda-Cuervo, Córdoba and Núñez-Antón [22], among others. Nevertheless, the choice of these parameters must be based on their adequacy to the specific application considered and its adverse effects on the posterior inference should be appropriately assessed and studied.…”

Section: Bayesian Estimationmentioning

confidence: 99%

“…Moreover, they highlighted the much shorter computation time that R-INLA required for the fitting of a model when compared to OpenBUGS. In Vranckx, Neyens and Faes [30], an extensive comparison of the fitting of the BYM model with OpenBUGS, CARBayes, R-INLA and other available software packages was also reported by fitting data from young people receiving diabetes medication in Belgian municipalities for the year 2014. Although no covariates were used in their analysis, the authors were able to identify locations with increased relative risk by fitting the BYM model to the data set under study.…”

Section: Introductionmentioning

confidence: 99%

Comparing Bayesian Spatial Conditional Overdispersion and the Besag–York–Mollié Models: Application to Infant Mortality Rates

Morales-Otero

Núñez‐Antón

2021

Mathematics

View full text Add to dashboard Cite

In this paper, we review overdispersed Bayesian generalized spatial conditional count data models. Their usefulness is illustrated with their application to infant mortality rates from Colombian regions and by comparing them with the widely used Besag–York–Mollié (BYM) models. These overdispersed models assume that excess of dispersion in the data may be partially caused from the possible spatial dependence existing among the different spatial units. Thus, specific regression structures are then proposed both for the conditional mean and for the dispersion parameter in the models, including covariates, as well as an assumed spatial neighborhood structure. We focus on the case of response variables following a Poisson distribution, specifically concentrating on the spatial generalized conditional normal overdispersion Poisson model. Models were fitted by making use of the Markov Chain Monte Carlo (MCMC) and Integrated Nested Laplace Approximation (INLA) algorithms in the specific context of Bayesian estimation methods.

show abstract

Comparison of different software implementations for spatial disease mapping

Cited by 10 publications

References 23 publications

A practical illustration of spatial smoothing methods for disconnected regions with INLA: spatial survey on overweight and obesity in Malaysia

A practical illustration of spatial smoothing methods for disconnected regions with INLA: spatial survey on overweight and obesity in Malaysia

High-resolution Bayesian mapping of landslide hazard with unobserved trigger event

Comparing Bayesian Spatial Conditional Overdispersion and the Besag–York–Mollié Models: Application to Infant Mortality Rates

Contact Info

Product

Resources

About