Fig. 2

Selection of variables and model hyperparameters. A Spearman or Pearson correlation matrix of continuous clinical variables. “ × ” means that the P value is less than 0.05, which is not significant. B Variable selection by using the Boruta algorithm C Seven variables were determined by the Boruta and Lasso algorithms. D–E Variable selection by using the Lasso regression. F–G Determination of optimal hyperparameters for the random forest model. H–I Determination of optimal hyperparameters for the neural network model. GCS glasgow coma scale, BE base excess, BMI body mass index, ISS injury severity score, Lasso least absolute shrinkage and selection operator