Table 1 Some related works focus on feature selection and medical classification.
From: Orthopedic disease classification based on breadth-first search algorithm
References | Major contribution | Methodology | Data | Enhancement |
|---|---|---|---|---|
Yao et al.5 | Patient prioritization in emergency department | Deep learning (DL) with CNN, RNN, attention | 864,043 patients over 5 years | 0.3–0.5% improvement in predicting death and admission |
Raita et al.6 | Emergency department triage | Machine learning models with triage data | 135,470 patients | Outperformed traditional ESI in predicting hospitalization, critical care, and mortality |
Kwon et al.7 | Emergency department patient outcome prediction | XGBoost with clinical data | Clinical data from the Korean National Emergency Department Information System (NEDIS) | Achieved high AUROC (0.93) and P-R (0.26) for predicting mortality, critical care, and admission |
Wang et al.8 | Hand X-ray hairline finger detection | Deep neural network (WrisNet) | 4346 hand X-rays | 0.05 improvement in average precision (AP) |
Pranata et al.9 | Calcaneus fracture detection in CT scans | Deep learning (ResNet vs. VGG) with SURF, canny edge detection, and contour tracing | Computed tomography (CT) images for calcaneus fractures | ResNet performed similar to VGG in accuracy but better with DNN |
Cheng et al.10 | Pelvic X-ray analysis | Deep learning network (PelviXNet) | 5204 pelvic X-rays | Achieved AUROC of 0.97, accuracy of 0.92, sensitivity of 0.90, and specificity of 0.93 |
Yaqoob et al.11 | Cancer classification with gene expression data | SCACSA (combines SCA & CSA algorithms) | Breast cancer datasets | Achieved superior accuracy compared to previous methods |
Joshi and Aziz12 | Disease classification | Hybrid method (CSSMO & SMOCS) with deep learning | 6 benchmark datasets | Outperformed deep learning and traditional machine learning models |
Mahto et al.14 | Disease classification | Feature selection (CSSMO) with deep learning | 8 cancer datasets | More accurate than existing machine learning and deep learning methods |
Saxena et al.15 | Feature selection for COVID-19 prediction | MPCA | Comparative analysis on standard datasets | Achieved high accuracy, used fewer features, and converged quickly |
Neggaz et al.16 | Feature selection | MRFOSCA (combines MRFO & SCA algorithms) | Public repository datasets | Achieved better accuracy, used fewer features, and worked well on large datasets |
Houssein et al.17 | ECG signal classification for heart disease diagnosis | MRFO-SVM | MIT-BIH database | Achieved good results in classifying ECG signals |
Hashim et al.18 | Feature selection for Parkinson’s disease diagnosis | mHGS (enhanced Hunger Games Search algorithm) | Voice recordings | Achieved better accuracy, used fewer features, and worked well for real-world tasks |
Hussain et al.19 | Feature selection | SCHHO | 16 datasets with varying dimensions | Reduced features by up to 87% while maintaining high accuracy |
