Algorithm 2: Model training and hyperparameter optimization for the CNN model.

1:Define ranges of model hyperparameter values to evaluate 2:for each external fold iteration do              ▹ External 5-fold CV 3:    Hold-out the samples of this fold (i.e., the test set) 4:    Use the rest of samples as calibration set 5:    for each iteration of the Hyperband algorithm do 6:        for each internal fold iteration in the calibration set do  ▹ Internal 5-fold CV 7:           Hold-out the samples of this fold (i.e., the validation set) 8:           Fit the model on the rest (i.e., the training set) 9:           Use the model to predict the held-out samples 10:           Calculate and store an accuracy metric for this fold 11:        end for 12:        Calculate the average performance across all folds 13:    end for 14:    Determine the optimal hyperparameter set 15:    Fit the final model using the calibration set with the optimal hyperparameters 16:    Use the model to predict the held-out samples 17:    Calculate and store accuracy metrics for this fold 18: end for 19:Calculate the average performance across all folds