diff --git a/README.Rmd b/README.Rmd index 2983b0f..b5d071a 100644 --- a/README.Rmd +++ b/README.Rmd @@ -26,17 +26,14 @@ See a vignette with NO2 exposure data: [[link]](https://changwoo-lee.github.io/b See bspme_1.0.0.pdf for the pdf file of the package manual. -**bspme** is an R package that provides a set of functions for **B**ayesian **sp**atial exposure **m**easurement **e**rror models, the Bayesian linear and generalized linear models with the presence of spatially correlated measurement error of covariate(s). For more details, please see the following paper: - -> Lee, C. J., Symanski, E., Rammah, A., Kang, D. H., Hopke, P. K., & Park, E. S. (2024). A scalable two-stage Bayesian approach accounting for exposure measurement error in environmental epidemiology. arXiv preprint arXiv:2401.00634. - - -The **bspme** package provides fast, scalable inference tools for Bayesian linear and generalized linear models with spatial exposure measurement error. +**bspme** is an R package that provides fast, scalable inference tools for **B**ayesian **sp**atial exposure **m**easurement **e**rror models, the Bayesian linear and generalized linear models with the presence of spatially correlated measurement error of covariate(s). These models typically arise from a two-stage Bayesian analysis of environmental exposures and health outcomes. From a first-stage model, predictions of the covariate of interest ("exposure") and their uncertainty information (typically contained in MCMC samples) are used to form a multivariate normal prior distribution $X\sim N(\mu, \Sigma)$ for exposure in a second-stage regression model. Naive, non-sparse choices of the precision matrix $Q = \Sigma^{-1}$ of the multivariate normal (such as a sample precision matrix) leads to the MCMC posterior inference algorithm infeasible to run for a large number of subjects $n$ because of the cubic computational cost associated with the $n$-dimensional MVN prior. -With the sparse precision matrix $Q$ obtained from the Vecchia approximation, the **bspme** package offers a fast, scalable algorithm to conduct posterior inference for large health datasets, with the number of subjects $n$ possibly reaching tens of thousands. +With a sparse precision matrix $Q$ obtained from the Vecchia approximation, the **bspme** package offers fast, scalable algorithms to conduct posterior inference with large health datasets, with the number of subjects $n$ possibly reaching tens of thousands. + For more details, please see the following paper: +> Lee, C. J., Symanski, E., Rammah, A., Kang, D. H., Hopke, P. K., & Park, E. S. (2024). A scalable two-stage Bayesian approach accounting for exposure measurement error in environmental epidemiology. arXiv preprint arXiv:2401.00634. ## Installation You can install the development version of bspme with the following code: diff --git a/README.md b/README.md index 6251135..4581e97 100644 --- a/README.md +++ b/README.md @@ -16,36 +16,31 @@ See a vignette with NO2 exposure data: See bspme_1.0.0.pdf for the pdf file of the package manual. -**bspme** is an R package that provides a set of functions for -**B**ayesian **sp**atial exposure **m**easurement **e**rror models, the -Bayesian linear and generalized linear models with the presence of -spatially correlated measurement error of covariate(s). For more -details, please see the following paper: +**bspme** is an R package that provides fast, scalable inference tools +for **B**ayesian **sp**atial exposure **m**easurement **e**rror models, +the Bayesian linear and generalized linear models with the presence of +spatially correlated measurement error of covariate(s). These models +typically arise from a two-stage Bayesian analysis of environmental +exposures and health outcomes. From a first-stage model, predictions of +the covariate of interest (“exposure”) and their uncertainty information +(typically contained in MCMC samples) are used to form a multivariate +normal prior distribution $X\sim N(\mu, \Sigma)$ for exposure in a +second-stage regression model. Naive, non-sparse choices of the +precision matrix $Q = \Sigma^{-1}$ of the multivariate normal (such as a +sample precision matrix) leads to the MCMC posterior inference algorithm +infeasible to run for a large number of subjects $n$ because of the +cubic computational cost associated with the $n$-dimensional MVN prior. +With a sparse precision matrix $Q$ obtained from the Vecchia +approximation, the **bspme** package offers fast, scalable algorithms to +conduct posterior inference with large health datasets, with the number +of subjects $n$ possibly reaching tens of thousands. For more details, +please see the following paper: > Lee, C. J., Symanski, E., Rammah, A., Kang, D. H., Hopke, P. K., & > Park, E. S. (2024). A scalable two-stage Bayesian approach accounting > for exposure measurement error in environmental epidemiology. arXiv -> preprint arXiv:2401.00634. - -The **bspme** package provides fast, scalable inference tools for -Bayesian linear and generalized linear models with spatial exposure -measurement error. These models typically arise from a two-stage -Bayesian analysis of environmental exposures and health outcomes. From a -first-stage model, predictions of the covariate of interest (“exposure”) -and their uncertainty information (typically contained in MCMC samples) -are used to form a multivariate normal prior distribution -$X\sim N(\mu, \Sigma)$ for exposure in a second-stage regression model. -Naive, non-sparse choices of the precision matrix $Q = \Sigma^{-1}$ of -the multivariate normal (such as a sample precision matrix) leads to the -MCMC posterior inference algorithm infeasible to run for a large number -of subjects $n$ because of the cubic computational cost associated with -the $n$-dimensional MVN prior. With the sparse precision matrix $Q$ -obtained from the Vecchia approximation, the **bspme** package offers a -fast, scalable algorithm to conduct posterior inference for large health -datasets, with the number of subjects $n$ possibly reaching tens of -thousands. - -## Installation +> preprint arXiv:2401.00634. \## +> Installation You can install the development version of bspme with the following code: