Figure 1.

Workflow of our analysis to select the best combination of features, optimize hyperparameters, and perform training of machine learning models. Starting from a set of the three features with the highest c-index, we increasingly add one feature at a time, optimize the hyperparameters with Optuna, train the model, and make a test validation. The train set and test set do not overlap, and repeated stratified k-folding with four splits is used to avoid overfitting of the models.