我尝试保存并稍后加载lightGBM模型，但无法执行此操作。我已经尝试了saveRDS（）和readRDS函数，但是当我预测时，我得到了这个错误：
预测器$predict（data = data，start_iteration = start_iteration，：尝试使用已不存在的Booster。如果您调用了Booster$finalize（）或使用saveRDS（）保存了此Booster，则可能会发生这种情况。要避免将来出现此错误，请使用saveRDS.lgb.Booster（）或Booster$save_model（）保存lightgbm Booster。
如果我在训练完模型后立即使用它，或者在将它加载到工作区的会话中使用它，它会工作得很好。问题是当我尝试在另一个应用程序上使用该模型时。下面是我如何编码的：

library(tidymodels)
library(lightgbm)
library(bonsai)

TMA_model <- data.frame(
  age = c(50, 45, 60, 55, 70, 34, 55, 48, 58, 42,
          52, 47, 62, 57, 72, 36, 53, 49, 59, 44,
          51, 46, 61, 56, 71, 35, 54, 50, 60, 43), 
  delta_creat = c(1.2, 1.0, 1.5, 1.3, 1.8, 1.2, 1.3, 1.4, 1.1, 1.6,
                  1.2, 1.0, 1.5, 1.3, 1.8, 1.2, 1.3, 1.4, 1.1, 1.6,
                  1.2, 1.0, 1.5, 1.3, 1.8, 1.2, 1.3, 1.4, 1.1, 1.6), 
  max_LDH = c(300, 280, 320, 310, 330, 295, 325, 290, 315, 305,
              305, 315, 290, 320, 300, 325, 280, 330, 310, 295,
              310, 295, 330, 280, 320, 305, 310, 325, 315, 290), 
  min_plat = c(150, 140, 160, 155, 170, 145, 165, 150, 135, 160,
               160, 155, 170, 145, 150, 160, 140, 170, 155, 145,
               155, 145, 170, 140, 160, 150, 155, 160, 165, 135),
  min_hb = c(12, 11.5, 13, 12.5, 14, 11.8, 13.2, 12.2, 12.6, 11.9,
             12.4, 11.7, 13.1, 12.3, 13.5, 11.6, 12.8, 12.1, 13.4, 12.0,
             12.9, 11.6, 13.2, 12.8, 13.0, 11.5, 12.7, 12.3, 12.5, 11.8), 
  max_ast = c(40, 38, 42, 41, 45, 37, 43, 39, 44, 40,
              42, 38, 44, 41, 40, 39, 43, 38, 45, 37,
              44, 37, 42, 38, 41, 40, 43, 39, 44, 38), 
  max_bt = c(37, 37.2, 37.5, 37.3, 37.8, 37.1, 37.4, 37.6, 37.0, 37.7,
             37.2, 37.1, 37.5, 37.4, 37.3, 37.0, 37.6, 37.8, 37.2, 37.4,
             37.7, 37.1, 37.3, 37.8, 37.0, 37.4, 37.2, 37.6, 37.5, 37.1), 
  max_ttap = c(100, 95, 105, 102, 110, 97, 108, 93, 103, 98,
               105, 100, 95, 110, 102, 108, 97, 93, 103, 95,
               100, 97, 105, 93, 108, 102, 110, 98, 103, 100), 
  max_tp = c(12, 12.2, 12.5, 12.3, 12.8, 12.1, 12.4, 12.6, 12.0, 12.7,
             12.3, 12.2, 12.1, 12.8, 12.4, 12.0, 12.6, 12.7, 12.5, 12.3,
             12.1, 12.4, 12.7, 12.8, 12.0, 12.2, 12.3, 12.6, 12.5, 12.4), 
  hypertension = c(1, 0, 1, 1, 0, 0, 1, 0, 1, 0,
                   1, 0, 1, 0, 1, 0, 0, 1, 1, 0,
                   0, 1, 0, 1, 0, 0, 1, 1, 0, 1),  
  MAP = c(90, 92, 88, 91, 89, 93, 87, 94, 86, 90,
          92, 94, 88, 90, 91, 89, 87, 93, 86, 92,
          91, 90, 93, 94, 86, 88, 89, 87, 92, 91),  
  TMA_class = factor(c(1, 2, 3, 4, 5, 1, 2, 3, 4, 5,
                       1, 2, 3, 4, 5, 1, 2, 3, 4, 5,
                       1, 2, 3, 4, 5, 1, 2, 3, 4, 5))
)

new_data <- data.frame(
  age = c(48, 55, 62, 54, 67),  
  delta_creat = c(1.2, 1.3, 1.1, 1.4, 1.6),  
  max_LDH = c(305, 290, 320, 310, 295),  
  min_plat = c(150, 160, 135, 155, 145),  
  min_hb = c(12.2, 12.6, 11.9, 12.4, 11.7),  
  max_ast = c(38, 43, 39, 44, 40),  
  max_bt = c(37.2, 37.4, 37.0, 37.6, 37.1),  
  max_ttap = c(95, 108, 93, 103, 98),  
  max_tp = c(12.2, 12.4, 12.0, 12.6, 12.7),  
  hypertension = c(1, 0, 1, 0, 1),  
  MAP = c(92, 87, 94, 86, 90)  
)

model_spec <- 
  boost_tree(mode = 'classification',
             mtry = 6,  
             trees = 50, 
             min_n = 50,  
             tree_depth = 5) %>%
  set_engine('lightgbm')

recipe <- recipe(TMA_class ~ ., data = TMA_model)

model_fit <- workflow() %>%
  add_recipe(recipe) %>%
  add_model(model_spec) %>%
  fit(data = TMA_model)

saveRDS(model_fit, "model_fit.rds")
model_b <- readRDS("model_fit.rds")

new_data_predictions <- predict(model_b, new_data)

print(new_data_predictions)

我在保存或加载它时没有错误，但我不能用加载的模型进行预测。还有，我对其他模型没有问题，像Lasso或XGBoost，它们工作得很好，但LightGBM没有。