使用brms实现多元Logistic Regression：响应变量相关性设置求助

我是brms的新用户，正在探索如何用brms进行多元Logistic回归。我拥有6个二分类响应变量，以及5个预测变量（1个连续变量、1个有序变量、3个二分类变量）。我知道可以使用bernoulli族进行建模，但当前的模型设定默认这6个响应变量相互独立，这与我使用多元回归的初衷相悖。我翻遍了brms的R文档也没找到替代方法，因此来向各位同行求助，万分感谢！

以下是复现问题的代码：

# Set seed for reproducibility
set.seed(123)

# Number of observations
n <- 1000

# Generate predictors
x1 <- rnorm(n, mean = 50, sd = 10)
x2 <- sample(1:5, n, replace = TRUE)
x3 <- rbinom(n, 1, 0.5)
x4 <- rbinom(n, 1, 0.3)
x5 <- rbinom(n, 1, 0.7)

# Generate response variables
y1 <- rbinom(n, 1, plogis(x1*0.02 + 0.3 + x2*0.1 + x3*0.5 + x4*0.4 + x5*0.6))
y2 <- rbinom(n, 1, plogis(x1*0.01 + 0.2 + x2*0.2 + x3*0.3 + x4*0.6 + x5*0.7))
y3 <- rbinom(n, 1, plogis(x1*0.03 + 0.1 + x2*0.3 + x3*0.4 + x4*0.5 + x5*0.2))
y4 <- rbinom(n, 1, plogis(x1*0.05 + 0.4 + x2*0.4 + x3*0.2 + x4*0.3 + x5*0.1))
y5 <- rbinom(n, 1, plogis(x1*0.02 + 0.6 + x2*0.1 + x3*0.7 + x4*0.8 + x5*0.9))
y6 <- rbinom(n, 1, plogis(x1*0.04 + 0.5 + x2*0.2 + x3*0.1 + x4*0.2 + x5*0.3))

# Combine predictors and response variables into a data frame
data <- data.frame(x1, x2, x3, x4, x5, y1, y2, y3, y4, y5, y6)

# brms model
model = brm(mvbind(y1,y2,y3,y4,y5,y6) ~ x1+x2+x3+x4+x5, data = data, family = bernoulli(link = "logit"))

系统信息 #

• 操作系统：Mac OS Monterey 12.4
• brms版本：2.19.0

内容的提问来源于stack exchange，提问作者some