Table 1.
Patient features and their domain.
Fig 1.
Henry gas solubility bases.
Fig 2.
Framework of the machine learning models trained and evaluated based one AQO with and without using HGSO to classify patient status outcome.
Fig 3.
The accuracy for (KNN, RF, C4.5, ANN, SVM and RF-SVM) classifiers with and without using HGSO.
Table 2.
The accuracy for (KNN, RF, C4.5, ANN, SVM and RF-SVM) classifiers based on HGSO.
Table 3.
Run parameters.
Table 4.
Comparison of classification model results.
Fig 4.
RF-SVM has the largest values of F-measure.
Fig 5.
RF-SVM based on AQO HGSO sensitivity.
Table 5.
Results of different parameters of ML classification models.
Fig 6.
Classification results for different number of pop_size (25–50) for each classifier employed: KNN, RF, C4.5, ANN, SVM, RF-SVM, without HGSO (a) and with HGSO (b).
(a) Results of Classification without HGSO. (b) Results of Classification with HGSO.
Table 6.
Comparative analysis of machine learning models with different pop sizes.
Fig 7.
Convergence curve of used algorithms with HGSO over the selected dataset.
Fig 8.
Convergence curve of used algorithms without HGSO over the selected dataset.