Fig. 3From: Application of machine learning in prediction of bone cement leakage during single-level thoracolumbar percutaneous vertebroplastyRandomforest screening variables. A The 21 most important factors calculated using the two random forest algorithms “%IncMSE” and “IncNodePurity.” B the use of the random forest selection model gradually increased the error when 1–10 variables were selected, and then the diagnostic error decreased graduallyBack to article page