关于BKMR model 使用时候遇见的问题

关于BKMR model 使用时候遇见的问题

BKMR （贝叶斯核函数回归）可以用于分析多种暴露物对结局的联合作用和影响
教程链接：
https://jenfb.github.io/bkmr/SimData1.html#3_fit_the_bkmr_model （for gaussian）

https://jenfb.github.io/bkmr/ProbitEx.html （for binomial）
When fitting probit BKMR, one can occasionally get the following error:

"Error in checkSymmetricPositiveDefinite(H, name = “H”) : H must be positive definite"

A solution that often works is to set the argument ‘est.h = TRUE’ in the kmbayes function. Doing this leads to a different sampler being used for one of the steps of the MCMC algorithm, which can get around this error.
参考文献：
Statistical software for analyzing the health effects of multiple concurrent exposures via Bayesian kernel machine regression. Jennifer F. Bobb, Birgit Claus Henn, Linda Valeri & Brent A. Coull Environmental Health 2018.