https://github.com/vanderschaarlab/survivalgan
SurvivalGAN: Generating Time-to-Event Data for Survival Analysis
https://github.com/vanderschaarlab/survivalgan
gan pytorch survival-analysis synthcity synthetic-data
Last synced: 6 months ago
JSON representation
SurvivalGAN: Generating Time-to-Event Data for Survival Analysis
- Host: GitHub
- URL: https://github.com/vanderschaarlab/survivalgan
- Owner: vanderschaarlab
- License: apache-2.0
- Created: 2023-01-23T09:48:30.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2023-02-27T15:46:18.000Z (over 2 years ago)
- Last Synced: 2025-04-05T07:23:02.982Z (6 months ago)
- Topics: gan, pytorch, survival-analysis, synthcity, synthetic-data
- Language: Jupyter Notebook
- Homepage:
- Size: 19.5 KB
- Stars: 27
- Watchers: 2
- Forks: 6
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# SurvivalGAN: Generating Time-to-Event Data for Survival Analysis
This repository contains the experimental code of SurvivalGAN, a generative model that handles survival data firstly by addressing the imbalance in the censoring and time horizons, and secondly by using a dedicated mechanism for approximating time-to-event/censoring. For more details, please read our AISTATS 2023 paper: 'SurvivalGAN: Generating time-to-event Data for Survival Analysis'.
The implementation of the method is included in the [synthcity library](https://github.com/vanderschaarlab/synthcity), in the [SurvivalGAN plugin](https://github.com/vanderschaarlab/synthcity/blob/main/src/synthcity/plugins/survival_analysis/plugin_survival_gan.py).
## Installation
Install `synthcity` and other depends
```bash
pip install -r requirements.txt
```
For more tutorials and examples, checkout the [Synthcity tutorials section](https://github.com/vanderschaarlab/synthcity#-tutorials).
## Datasets
Add the data in the `experiments/data` folder.
| Dataset | No. instances | No. censored instances | No. features | Access |
|---------------------------------|------------------------|---------------------------------|-----------------------|---------------------------|
| ACTG 320 clinical trial dataset | 1151 | 1055 | 11 | [Link](https://github.com/sebp/scikit-survival/blob/master/sksurv/datasets/data/actg320.arff) |
| METABRIC | 1093 | 609 | 689 | [Link](https://www.kaggle.com/datasets/raghadalharbi/breast-cancer-gene-expression-profiles-metabric) |
| CUTRACT | 10086 | 8881 | 6 | private |
| PHEART | 40409 | 25664 | 29 | private |
| SEER prostate cancer | 171942 | 167568 | 6 | private |
## Reproducing results
| **Result** | **Source notebook** |
|-------------------|------------------------------------------------------------------------------------------------------------------------|
| Figure 1 | [experiments_00_km_plots_tte_models](experiments/experiments_00_km_plots_tte_models.ipynb) |
| Table 1,2,9,10,15 | [experiments_01_benchmark_synthetic_survival_data](experiments/experiments_01_benchmark_synthetic_survival_data.ipynb) |
| Table 3 | [experiments_02_sources_of_gain_parametric](experiments/experiments_02_sources_of_gain_parametric.ipynb) |
| Table 11 | [experiments_04_loglikelihood](experiments/experiments_04_loglikelihood.ipynb) |
| Table 12, 13, 14 | [experiments_05_predicting_censoring](experiments/experiments_05_predicting_censoring.ipynb) |
| Table 16 | [experiments_03_gmm_test_perf](experiments/experiments_03_gmm_test_perf.ipynb) |
| Figure 4,5,8 | [plots_00_data_fidelity](experiments/plots_00_data_fidelity.ipynb) |
| Figure 6,7 | [plots_02_benchmark_gain_of_function](experiments/plots_02_benchmark_gain_of_function.ipynb) |
## Citing
```
@misc{https://doi.org/10.48550/arxiv.2302.12749,
doi = {10.48550/ARXIV.2302.12749},
url = {https://arxiv.org/abs/2302.12749},
author = {Norcliffe, Alexander and Cebere, Bogdan and Imrie, Fergus and Lio, Pietro and van der Schaar, Mihaela},
keywords = {Machine Learning (cs.LG), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {SurvivalGAN: Generating Time-to-Event Data for Survival Analysis},
year = {2023},
copyright = {Creative Commons Attribution 4.0 International}
}
```