Approximate Bayesian Inference for Latent Gaussian models by using Integrated Nested Laplace Approximations

Rue, Håvard; Martino, Sara; Chopin, Nicolás

doi:10.1111/j.1467-9868.2008.00700.x

Cited by 4,208 publications

(4,713 citation statements)

References 168 publications

(192 reference statements)

Supporting

Mentioning

4,451

Contrasting

Unclassified

Order By: Relevance

“…Covariates were given normally distributed, uninformative priors with a precision of 0.001, while spatial errors were fitted with log‐gamma priors and a precision of 0.005. Analyses were carried out in R v3.2.2 using the package R‐INLA (Rue, Martino & Chopin, 2009). …”

Section: Methodsmentioning

confidence: 99%

Does functional homogenization accompany taxonomic homogenization of British birds and how do biotic factors and climate affect these processes?

White

Montgomery

Storchová

et al. 2018

Ecology and Evolution

View full text Add to dashboard Cite

Environmental change has reshuffled communities often causing taxonomic homogenization rather than differentiation. Some studies suggest that this increasing similarity of species composition between communities is accompanied by an increase in similarity of trait composition—functional homogenization—although different methodologies have failed to come to any consistent conclusions. Functional homogenization could have a large effect on ecosystem functioning and stability. Here, we use the general definition of homogenization as “reduced spatial turnover over time” to compare changes in Simpson's beta diversity (taxonomic turnover) with changes in Rao's quadratic entropy beta diversity (functional turnover) in British breeding birds at three spatial scales. Using biotic and climatic variables, we identify which factors may predispose a site to homogenization. The change in turnover measures between two time periods, 20 years apart, was calculated. A null model approach was taken to identify occurrences of functional homogenization and differentiation independent of changes in taxonomic turnover. We used conditional autoregressive models fitted using integrated nested Laplace approximations to determine how environmental drivers and factors relating to species distributions affect changes in spatial turnover of species and functional diversity. The measurement of functional homogenization affects the chance of rejection of the null models, with many sites showing taxonomic homogenization unaccompanied by functional homogenization, although occurrence varies with spatial scale. At the smallest scale, while temperature‐related variables drive changes in taxonomic turnover, changes in functional turnover are associated with variation in growing degree days; however, changes in functional turnover become more difficult to predict at larger spatial scales. Our results highlight the multifactorial processes underlying taxonomic and functional homogenization and that redundancy in species traits may allow ecosystem functioning to be maintained in some areas despite changes in species composition.

show abstract

Section: Methodsmentioning

confidence: 99%

Does functional homogenization accompany taxonomic homogenization of British birds and how do biotic factors and climate affect these processes?

White

Montgomery

Storchová

et al. 2018

Ecology and Evolution

View full text Add to dashboard Cite

show abstract

“…These data were used within a Bayesian hierarchical space-time model implemented through an adapted Stochastic Partial Differential Equations (SPDE) approach using Integrated Nested Laplace Approximations (INLA) for inference [32,33] to develop a 1 × 1km gridded prediction map of parasite prevalence among children aged 2 to 10 years ( Pf PR 2-10 ) projected to the year 2015. The model adjusted for a minimal set of conservative, long-term covariates traditionally used in vector-borne diseases.…”

Section: Methodsmentioning

confidence: 99%

Coverage of routine reporting on malaria parasitological testing in Kenya, 2015–2016

Maina

Macharia

Ouma

et al. 2017

Global Health Action

View full text Add to dashboard Cite

Background: Following the launch of District Health Information System 2 across facilities in Kenya, more health facilities are now capable of carrying out malaria parasitological testing and reporting data as part of routine health information systems, improving the potential value of routine data for accurate and timely tracking of rapidly changing disease epidemiology at fine spatial resolutions. Objectives: This study evaluates the current coverage and completeness of reported malaria parasitological testing data in DHIS2 specifically looking at patterns in geographic coverage of public health facilities in Kenya. Methods: Monthly facility level data on malaria parasitological testing were extracted from Kenya DHIS2 between November 2015 and October 2016. DHIS2 public facilities were matched to a geo-coded master facility list to obtain coordinates. Coverage was defined as the geographic distribution of facilities reporting any data by region. Completeness of reporting was defined as the percentage of facilities reporting any data for the whole 12-month period or for 3, 6 and 9 months. Results: Public health facilities were 5,933 (59%) of 10,090 extracted. Fifty-nine per Cent of the public facilities did not report any data while 36, 29 and 22% facilities had data reported at least 3, 6 and 9 months, respectively. Only 8% of public facilities had data reported for every month. There were proportionately more hospitals (86%) than health centres (76%) and dispensaries/clinics (30%) reporting. There were significant geographic variations in reporting rates. Counties along the malaria endemic coast had the lowest reporting rate with only 1% of facilities reporting consistently for 12 months. Conclusion: Current coverage and completeness of reporting of malaria parasitological diagnosis across Kenya’s public health system remains poor. The usefulness of routine data to improve our understanding of sub-national heterogeneity across Kenya would require significant improvements to the consistency and coverage of data captured by DHIS2.

show abstract

“…Finally using the GMRF representation, it carries out computation taking advantage of sparse matrices. Rue et al (2009) proposed an efficient integrated nested Laplace approximation (INLA). The first step of INLA is an approximationp(θ|y) of marginal posterior of θ:…”

Section: Markov Random Field Approximationmentioning

confidence: 99%

Review on statistical methods for large spatial Gaussian data

Park¹

2015

Journal of the Korean Data and Information Science Society

View full text Add to dashboard Cite

The Gaussian geostatistical model has been widely used for modeling spatial data. However, this model suffers from a severe difficulty in computation because inference requires to invert a large covariance matrix in evaluating log-likelihood. In addressing this computational challenge, three strategies have been employed: likelihood approximation, lower dimensional space approximation, and Markov random field approximation. In this paper, we reviewed statistical approaches attacking the computational challenge. As an illustration, we also applied integrated nested Laplace approximation (INLA) technology, one of Markov approximation approach, to real data to provide an example of its use in practice dealing with large spatial data.

show abstract

Approximate Bayesian Inference for Latent Gaussian models by using Integrated Nested Laplace Approximations

Cited by 4,208 publications

References 168 publications

Does functional homogenization accompany taxonomic homogenization of British birds and how do biotic factors and climate affect these processes?

Does functional homogenization accompany taxonomic homogenization of British birds and how do biotic factors and climate affect these processes?

Coverage of routine reporting on malaria parasitological testing in Kenya, 2015–2016

Review on statistical methods for large spatial Gaussian data

Contact Info

Product

Resources

About