How marcuschong2003/chbmit works
This is a collection of research scripts for EEG seizure detection using CHB-MIT data, plus a separate Bonn-dataset SVM experiment. The main workflow parses seizure timings, filters and segments EEG into ictal/non-ictal CSV data, then trains and evaluates CNN classifiers. The repository also contains patient-wise CNN experiments and a Bonn feature-based SVM workflow. README documentation is minimal; the map reflects the sampled scripts and the file tree, not an integrated application.
Main entry points: attention_independent.py Patient-wise attention CNN · generate_csv.py CSV generator · data_model.py Pooled CNN · attention_model.py Pooled attention CNN
Components 14
EEG preprocessing
- CSV generator
generate_csv.py - Ictal / non-ictal CSVs
CHB-MIT models
- Pooled CNN
data_model.py - Pooled attention CNN
attention_model.py - Patient-wise CNN
patient_independent_model.py - Patient-wise attention CNN
attention_independent.py - Model evaluation
- Saved attention model
Bonn experiment
- Bonn SVM experiment
Conference/Bonn/10-fold CV SVM.py - SVM metrics
Other components
- Researcher
- CHB-MIT EEG
- Patient CSVs
- Bonn formatted CSV
Connections 19
- Researcher → CSV generator: runs preprocessing
- CHB-MIT EEG → CSV generator: reads EEG and summaries
- CSV generator → Ictal / non-ictal CSVs: writes segments
- Researcher → Pooled CNN: runs training
- Ictal / non-ictal CSVs → Pooled CNN: reads samples
- Pooled CNN → Model evaluation: reports evaluation
- Researcher → Pooled attention CNN: runs training
- Ictal / non-ictal CSVs → Pooled attention CNN: reads samples
- Pooled attention CNN → Model evaluation: reports evaluation
- Researcher → Patient-wise CNN: runs training
- Patient CSVs → Patient-wise CNN: reads patient data
- Patient-wise CNN → Model evaluation: reports evaluation
- Researcher → Patient-wise attention CNN: runs training
- Patient CSVs → Patient-wise attention CNN: reads patient data
- Patient-wise attention CNN → Model evaluation: reports evaluation
- Patient-wise attention CNN → Saved attention model: saves model
- Researcher → Bonn SVM experiment: runs experiment
- Bonn formatted CSV → Bonn SVM experiment: reads samples
- Bonn SVM experiment → SVM metrics: reports metrics