This repository is for the paper entitled “Predicting seizure outcome after epilepsy surgery: do we need more complex models, larger samples, or better data?”.
Preprint: https://www.medrxiv.org/content/10.1101/2023.02.13.23285866v1
Epilepsia publication: https://onlinelibrary.wiley.com/doi/10.1111/epi.17637
- Data overview, describing what data were included in the paper and how these data were categorised
- Jupyter notebooks for running the paper's logistic regression, multilayer perceptron, and XGBoost models
- Jupyter notebooks and R scripts for creating the paper's figures
- 00-LR-model.ipynb* - Jupyter notebook for the paper's logistic regression model.
- 01-MLP-model.ipynb* - Jupyter notebook for the paper's multilayer perceptron model.
- 02-XGBoost-own-model.ipynb* - Jupyter notebook for the paper's XGBoost model (that was trained on our data).
- 03-Figure-2.R - R script used for producing Figure 2 in the paper.
- 04-Figures-3A-3B-4A.ipynb - Jupyter notebook for producing Figure 3A, Figure 3B, and Figure 4A in the paper.
- 05-Figure-4B.R - R script used for producing Figure 4B in the paper.
*Models require prior selection of predictor variables, one-hot encoding of categorical predictor variables, and handling of missing data.
Questions/suggestions/feedback would be much appreciated!