Table 2 Summary of Related Work on EEG-based Alzheimer’s Disease Detection.
Study | Preprocessing techniques | Feature extraction | Classifier | Explainability | Key limitation |
|---|---|---|---|---|---|
Band-pass filtering, segmentation | Spectral, statistical descriptors | SVM, DT | No | Shallow learning only | |
ASR, rereferencing, ICA | Entropy, fractal, connectivity | KNN, RF, LightGBM | No | No interpretability framework | |
ICA, normalization | Connectivity features | DT, XGBoost | Partial (ROC only) | No SHAP or PC-based insights | |
Minimal processing | Time-frequency + CNN | CNN, FNN | No | Deep-only, lacks fusion | |
Basic filtering | Deep learned features | CNN | No | No handcrafted fusion | |
Varies | Hybrid (deep + handcrafted) | Ensemble models | Partial (AUC, ROC) | Lacks SHAP or medical interpretation |
