Missing data imputation in machine learning pipelines

Machine learning is an important part of working in R. Packages like mlr3 simplify the whole process. Its no need to manually split data into training and test set, no need to manually fit linear models. Even more, packages like mlr3pipelines let you crate complex pipelines and include feature engineering and transformation in your models. R is also used by statisticians, from statisticians we have advanced methods of imputing missing data like mice or Amelia. What happens when we want to connect machine learning with a statistical approach.