Predicting clinical status of patients after an acute ischemic stroke using random forests