使用 R 中的 MCMC function 计算置信区间

Question

我想计算使用马尔可夫链蒙特卡罗模拟估计的估计相对风险列表的 CI。 我在 R 中使用 SpatialEpi package 中的贝叶斯聚类检测方法。 此 function 返回空间聚类检测中包含的每个“区域”的估计相对风险列表。 以下是如何计算 RR 的原始代码：

    process_MCMC_sample <-
          function(sample, param, RR.area, cluster.list, cutoffs){

           n <- length(RR.area)
           n.sim <- length(sample)

         # Store outputs here. These are averages weighted by the probability of 
         # each configuration
         high.area <- low.area <- RR.est.area <- rep(0, n)

           # Zone based measures
          high.z <- pgamma(cutoffs$high, param$shape, param$rate, lower.tail=FALSE)
          low.z <- pgamma(cutoffs$low, param$shape, param$rate, lower.tail=TRUE)
          RR.z <- param$shape / param$rate


   #-------------------------------------------------------------------------------
   # Find probabilities of cluster and anti-cluster membership & estimated 
   # relative risk
   #-------------------------------------------------------------------------------
   # Get probabilities of each configuration
   unique <- unique(sample)
   indices <- match(sample, unique)
   prob <- table(indices)/n.sim
   prob.names <- as.numeric(names(prob))

     # Over all sampled configurations
      for(i in 1:length(prob.names)){
         config <- unique[[prob.names[i]]]
          k <- length(config)

         if(k==0){
         RR.est.area <- RR.est.area + RR.area * prob[i]
        }else{
         # Used to identify all areas included in configuration
         config.areas <- NULL    
          for(j in 1:k){        
       # All areas in this single zone in this configuration
        zone <- config[j]
        areas <- cluster.list[[zone]]
        config.areas <- c(config.areas, areas)
  
     # Update all areas inside configuration
       high.area[areas] <- high.area[areas] + high.z[zone] * prob[i]
       low.area[areas] <- low.area[areas] + low.z[zone] * prob[i]
       RR.est.area[areas] <- RR.est.area[areas] + RR.z[zone] * prob[i] 
      }

       # Update all areas outside configuration
       outside.areas <- setdiff(1:n, config.areas)
       RR.est.area[outside.areas] <- 
       RR.est.area[outside.areas] + RR.area[outside.areas]*prob[i]
     }
    }

      return(list(
      high.area=high.area,
      low.area=low.area,
      RR.est.area=RR.est.area)
      )
      }

我想在其中添加代码来估计每个估计的相对风险的 95% CI。