Table 3.
EEG healthcare applications enabled by deep learning.
| Category | Application domain | EEG paradigm | Deep learning models | Key outcomes | Example datasets |
|---|---|---|---|---|---|
| Clinical Diagnosis | Neurological Disorder Diagnosis (Epilepsy, AD, PD) | Resting EEG, Task EEG | CNN, LSTM, CNN–LSTM, Transformer, GNN | >90% accuracy in seizure detection; early biomarkers for AD/PD identified | Bonn, CHB-MIT |
| Functional Monitoring | Brain–Computer Interfaces (BCIs) | Motor Imagery, SSVEP, P300 | CNN, CNN–LSTM, Transformer | High information transfer rate (ITR); real-time decoding; robust cross-subject performance |
BCI Competition IV |
| Sleep & Fatigue Monitoring | Polysomnography EEG | CNN, SeqSleepNet, Transformer | Automated sleep staging (>85% accuracy); real-time fatigue detection | Sleep-EDF | |
| Mental Health | Emotion & Mental Disorder Assessment | Resting/Task EEG | CNN, LSTM, Transformer, GNN | High accuracy in depression and emotion detection; interpretable biomarkers for psychiatric disorders | DEAP, SEED |
| Rehabilitation & Therapy | Motor Rehabilitation & Neurofeedback | Motor Imagery EEG | CNN, RNN, GCN, Attention-enhanced Transformers | Improved motor recovery in stroke; adaptive neurofeedback enhances cognitive performance | Custom clinical datasets |