Fig 1.
The methodological analysis of the proposed research study.
Table 1.
The maternal health dataset features related information.
Table 2.
The maternal health dataset features related information.
Fig 2.
The dataset features distribution analysis concerning correlation to risk levels.
Fig 3.
Statistical analysis of dataset features for correlation.
Fig 4.
3D analysis of prominent features concerning maternal pregnancy risk level and age.
Fig 5.
The pair plot analysis of dataset features.
Fig 6.
Number of samples for the dataset, (a) Before resampling, and (b) After applying SMOTE for resampling.
Fig 7.
Graphical illustration of hybrid feature set using proposed DT-BiLTCN.
Fig 8.
The maternal health risk analysis and BiLTCN model risk prediction.
Table 3.
The proposed BiLTCN model configuration parameters.
Fig 9.
The architectural layers analysis of the BiLTCN approach.
Table 4.
The proposed model compilation parameters with hyperparameter tuning.
Table 5.
Comparative analysis of accuracy score of employed approaches based on original dataset features with imbalanced data.
Table 6.
Comparative analysis of the accuracy of employed approaches based on original dataset features using SMOTE-balanced data.
Table 7.
Training accuracy and loss of BiLTCN concerning the number of iterations.
Table 8.
Results for the BiLTCN for all classes.
Table 9.
BiLTCN model results with hyperparameter tuning.
Fig 10.
ROC curve accuracy analysis of BiLTCN.
Fig 11.
BiLTCN approach performance analysis.
Table 10.
Averaged results of models using the proposed DT-BiLTCN feature extraction technique with the imbalanced dataset.
Table 11.
Averaged results of applied learning models using the proposed DT-BiLTCN feature extraction technique.
Table 12.
SVM category-wise classification report.
Table 13.
Comparative analysis of DT and BiLTCN as feature extraction techniques.
Fig 12.
Comparative analysis of proposed DT-BiLTCN feature extraction with DT and BiLTCN.
Table 14.
The 10-fold cross-validation analysis of employed approaches with proposed DT-BiLTCN technique.
Fig 13.
The feature space analysis is based on class distribution using the original features and class distribution using the proposed DTBiLTCN.
Table 15.
The computational cost analysis (time in seconds) of machine learning methods.
Table 16.
Performance analysis with existing approaches.