Question

我正在分析一项针对临床药物和酒精试验参与者戒酒率的实验。有两组，接受新疗法的人群和接受安慰剂的人群。前一周的禁欲在0（基线），4、8、12和24周进行。损耗/丢失数据的发生率很高，因此在开始进行试验的128个数据中，只有55个在随访时提供了数据。

这里是数据。

df <- data.frame(id = factor(rep(1:128,each=5)),
                 time = rep(c(0,4,8,12,24),times=128),
                 group = c(rep("placebo",335),rep("treatment",305)),
                 abs = c(0, 0, 0, 0, 0, 0, NA, NA, NA, NA, 0, NA, NA, NA, NA, 0, 1, 1, NA, 1, 0, 0, 0, NA, NA, 0, 0, 1, NA, 0, 0, 0, NA, NA, 1, 0, NA, NA, NA, 0, 0, 0, NA, NA, NA, 0, NA, NA, NA, 1, 0, 0, NA, NA, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, NA, NA, NA, 0, NA, NA, NA, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, NA, NA, 0, 0, NA, NA, NA, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, NA, 0, 0, 0, 0, NA, 0, NA, NA, NA, NA, 0, 0, NA, NA, NA, 0, 0, NA, 0, 0, 0, 0, 0, 0, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, NA, 0, 0, NA, NA, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, NA, 0, 0, NA, 0, 0, 0, 0, 0, NA, 0, 0, 0, 0, 0, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, NA, 0, NA, NA, NA, NA, 0, 0, 0, NA, NA, 0, 1, 1, 1, NA, 0, 0, 0, 0, NA, 0, 0, 0, NA, NA, 0, 0, 0, 0, NA, 0, 0, 0, 0, 0, 0, NA, NA, NA, NA, 0, 0, NA, NA, NA, 0, 0, 1, 1, NA, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, NA, 0, 0, 1, 1, 1, 0, 0, 0, NA, NA, 0, 0, 0, 0, NA, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, NA, 0, 0, NA, NA, NA, 0, 0, 0, 0, 0, 0, NA, NA, NA, NA, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, NA, NA, NA, 0, NA, NA, NA, NA, 0, 0, 0, 0, NA, 0, 0, 0, 0, NA, 0, NA, NA, NA, NA, 0, 0, 0, 0, 1, 0, 1, NA, NA, 0, 0, 0, 0, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, NA, NA, 0, NA, NA, NA, NA, 0, NA, NA, NA, NA, 0, NA, NA, NA, NA, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, NA, 0, NA, 0, NA, NA, NA, NA, 0, 0, 0, 0, 0, 0, 0, NA, NA, NA, 0, 0, NA, 0, NA, 0, NA, 1, NA, 1, 0, 0, NA, NA, NA, 0, 1, 1, 1, 1, 0, 0, 0, NA, NA, 0, NA, NA, NA, NA, 0, NA, NA, NA, NA, 0, 0, 0, 0, NA, 0, 0, 0, 0, NA, 0, 0, NA, 0, NA, 0, 0, 0, 0, 0, 0, NA, NA, NA, NA, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, NA, NA, NA, NA, 0, 0, NA, NA, NA, 0, 0, 0, 0, NA, 0, 0, NA, 0, NA, 0, 0, 0, 0, NA, 0, 0, 0, 0, NA, 0, 0, 0, 0, NA, 0, NA, NA, NA, NA, 0, 1, 1, 1, 1, 0, NA, NA, NA, NA, 0, 0, NA, NA, NA, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, NA, NA, NA, 0, NA, NA, NA, NA, 0, 0, 0, 0, 1, 0, 0, NA, NA, NA, 0, 0, 0, 0, 0, 0, 0, NA, NA, NA, 0, 1, 1, 1, 1, 0, 0, 1, 0, NA, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, NA, 0, NA, 0, 0, NA, 0, NA, 0, 0, 0, 0, NA, 0, 0, NA, NA, NA))

随着时间的流逝，观察到的病例的比例看起来像这样

从表面上看，这看起来像是不错的治疗效果，尤其是在第24周时。

我在mixed_model包中使用了GLMMadaptive函数来运行纵向逻辑回归模型，因为该包允许计算总体平均系数。现在，当我运行将时间作为连续变量的模型时，模型可以很好地收敛

library(GLMMadaptive)
m <- mixed_model(abs ~ group*time,
                 random = ~1|id,
                 data = df,
                 family = binomial())
summary(m)

#   model output

#   Call:
#   mixed_model(fixed = abs ~ group * time, random = ~1 | id, data = df, 
#               family = binomial())
# 
#   Data Descriptives:
#   Number of Observations: 440
#   Number of Groups: 128 
# 
#   Model:
#   family: binomial
#   link: logit 
# 
#   Fit statistics:
#   log.Lik AIC BIC
#   -137 285 299
# 
#   Random effects covariance matrix:
#   StdDev
#   (Intercept)   2.68
# 
#   Fixed effects:
#                       Estimate Std.Err z-value p-value
#   (Intercept)          -4.9350  0.8431  -5.854  <1e-04
#   grouptreatment        0.0913  0.8971   0.102   0.919
#   time                  0.1070  0.0362   2.956   0.003
#   grouptreatment:time   0.0952  0.0530   1.798   0.072
# 
#   Integration:
#   method: adaptive Gauss-Hermite quadrature rule
#   quadrature points: 11
# 
#   Optimization:
#   method: hybrid EM and quasi-Newton
#   converged: TRUE

但是，我真正想做的是在每个时间点进行对比计算，计算出禁食几率的组差异，并控制其他时间点。我对第24周戒酒几率的差异特别感兴趣（控制所有其他时间点）。所以我想把时间当作一个因素。但是，当我运行 this 模型时：

m <- mixed_model(abs ~ group*factor(time),
                 random = ~1|id,
                 data = df,
                 family = binomial())
summary(m)

它无法收敛，我收到以下错误消息。

Error in mixed_fit(y, X, Z, X_zi, Z_zi, id, offset, offset_zi, family,  : 
                     A large coefficient value has been detected during the optimization.
                   Please re-scale you covariates and/or try setting the control argument
                   'iter_EM = 0'. Alternatively, this may due to a
                   divergence of the optimization algorithm, indicating that an overly
                   complex model is fitted to the data. For example, this could be
                   caused when including random-effects terms (e.g., in the
                                                               zero-inflated part) that you do not need. Otherwise, adjust the
                   'max_coef_value' control argument.

我一直遵循此错误消息中的建议以及软件包here的作者提供的关于优化的建议，但均无济于事。但是，我感觉就像是狒狒试图指挥维也纳爱乐乐团，结果是您所期望的，全部都是随机的，有点混乱。

该模型不适合吗？如果没有，没有人有任何提示如何使它收敛？如果是这样，那么有人对其他方法有任何建议吗？

GLMM支持包中的纵向逻辑回归模型的收敛性问题

0 个答案: