---
language:
- en
license: odc-by
tags:
- medical
- eeg
- tabular-classification
- seizure-prediction
metrics:
- accuracy
- f1
- roc_auc
---

# Neuro Pulse Model

This repository contains trained models for predicting epileptic seizures using EEG feature data derived from the CHB-MIT Scalp EEG Database.

Research prototype. Not a medical device; not for clinical decisions.

Code: https://github.com/JayD108/Neuro_Pulse

> **Data provenance:** The EEG feature data in this repository is derived from the CHB-MIT Scalp EEG Database v1.0.0, distributed by PhysioNet under the Open Data Commons Attribution License v1.0 (ODC-By 1.0). This repository does not contain or redistribute the original `.edf` recordings. See the [License / Data Provenance](#license--data-provenance) section below.

## Models
- models/trained/apala_xgb_v1.json: XGBoost (500 trees, depth 6) trained on all 24 patients, 590 features,
  XGBoost native format (no pickle). Used by the APALA dashboard.
  Leave-one-patient-out ROC-AUC: mean 0.56, median 0.59, range 0.26–0.71 (per patient: apala_xgb_v1_lopo.json).
- models/trained/xgboost_model.pkl: original V1 XGBoost, 180 features, trained on chb01 only.
  ROC-AUC 0.998 on chb01 itself; it does not generalise to other patients.
- master_model.pkl / ensemble_model.pkl: V1 stacking ensembles (research). Need scikit-learn to load.

## Data Information

The processed CSVs are in the dataset repo https://huggingface.co/datasets/JayF14/chb-mit-preprocessed-eeg, not in this repo.

- The features are provided in CSV format (`chbXX_labeled_features.csv`).
- The original raw `.edf` EEG recordings are **not** included here; they are available from PhysioNet: https://physionet.org/content/chbmit/1.0.0/

## How to Use

Load the APALA model:

```python
from huggingface_hub import hf_hub_download
import xgboost as xgb

path = hf_hub_download("JayF14/Neuro_Pulse", "models/trained/apala_xgb_v1.json")
model = xgb.XGBClassifier()
model.load_model(path)
```

## License / Data Provenance

The EEG feature data in this repository is a **processed/feature-engineered derivative** of the **CHB-MIT Scalp EEG Database v1.0.0**.

- **Original dataset:** CHB-MIT Scalp EEG Database v1.0.0
- **PhysioNet page:** https://physionet.org/content/chbmit/1.0.0/
- **DOI:** 10.13026/C2K01R
- **Source license:** Open Data Commons Attribution License v1.0 (ODC-By 1.0)
  - License text: https://opendatacommons.org/licenses/by/1.0/

The creator of this repository does **not** claim ownership of the underlying EEG recordings. The original CHB-MIT EEG recordings are distributed by PhysioNet under the Open Data Commons Attribution License v1.0.

When using or redistributing this work, please include the following attribution:

> This work uses data derived from the CHB-MIT Scalp EEG Database v1.0.0 (Guttag, 2010), distributed by PhysioNet (https://physionet.org/content/chbmit/1.0.0/) under the Open Data Commons Attribution License v1.0 (https://opendatacommons.org/licenses/by/1.0/). DOI: 10.13026/C2K01R.

## Citation

```bibtex
@article{guttag2010chbmit,
  author  = {Guttag, John},
  title   = {{CHB-MIT Scalp EEG Database}},
  journal = {{PhysioNet}},
  year    = {2010},
  month   = jun,
  note    = {Version 1.0.0},
  doi     = {10.13026/C2K01R},
  url     = {https://doi.org/10.13026/C2K01R}
}
```