Table 4.
Consolidated list of ML algorithms and XAI methods from the included studies in rehabilitation.
| Subsection | Author(s) | ML algorithm used | XAI method used |
|---|---|---|---|
| Rehabilitation | Gandolfi et al. (2023) | Random Forest regression | SHAP, LIME, PFI, RFI |
| Lee and Choy (2023) | Feed-forward neural network | Saliency map (plus gradient-based and threshold-based methods) | |
| Lee et al. (2024) | Multiple compared; Random Forest best | Permutation importance, SHAP | |
| Abdelaziz et al. (2025) | Random Forest, SVM | Anchors | |
| Gupta et al. (2025) | Compared RF, SVM, LR, DT, KNN (SHAP on optimized KNN) | SHAP | |
| Hussain and Jany (2024) | Gradient Boosting, Histogram Gradient Boosting, Random Forest | SHAP (TreeSHAP), LIME, Anchors | |
| Slijepcevic et al. (2024) | Linear SVM (best), also RF, CNN, MLP, DT | LRP | |
| Slijepcevic et al. (2023) | CNNs, SNNs, RFs, DTs (RF best for certain angles) | Grad-CAM (1D), Gini feature importance | |
| Guo et al. (2024) | Stacking model and LDA, DT, GBC, LightGBM, ETC, XGBoost, LR, RF, KNN | SHAP | |
| He et al. (2024) | LDA, DT, GBC, LightGBM, ETC, XGBoost, LR, RF | SHAP |