Fig 1.
Overall research workflow.
Fig 2.
A diagram illustration the neural network structure for feature selection.
Fig 3.
AUC values for all tested feature selection model and classification model combinations.
Fig 4.
A box plot to display the AUC value range for each feature selection model with seven classification models for each feature category.
Fig 5.
Feature selection frequency in RF + KSVM.
3 features selected in each fold of LOOCV.
Fig 6.
Feature selection frequency in RF + NN.
2 features selected in each fold of LOOCV.
Fig 7.
ROC curves for RF+KSVM and RF+NN combinations.
Both AUC = 0.889.
Table 1.
Comparison of AUCs with/without machine learning feature selection for delta-features.