Discrete Time Survival Analysis
PyDTS is a Python package designed for discrete-time survival analysis with competing risks, offering tools for model fitting, evaluation, and simulation.
For details, usage examples, and API information - check out the package documentation
PyDTS offers:
- Discrete-time competing-risks regression models.
- Automated procedures for hyperparameter tuning.
- Sure Independence Screening methods for feature selection.
- Model evaluation metrics for predictive accuracy and calibration.
- Simulation tools for generating synthetic datasets for research and testing.
Additional simulations and illustrative examples are available in Meir and Gorfine (2025), Discrete-Time Competing-Risks Regression with or without Penalization, Biometrics (2025), and in the accompanying Github Repository
Installation
PyDTS can be installed using PyPI as follows:
pip install pydts
Dependencies
PyDTS supports Python versions 3.9–3.13.
The package requires the following dependencies (with version constraints chosen for compatibility across Python and NumPy/SciPy releases):
-
NumPy
- Python 3.9:
>=1.26,<2.1 - Python 3.10:
>=1.26,<2.3 - Python 3.11–3.13:
>=1.26(including NumPy 2.x)
- Python 3.9:
-
SciPy
- Python 3.9:
>=1.13,<1.14 - Python 3.10:
>=1.14,<1.16 - Python 3.11–3.13:
>=1.15
- Python 3.9:
-
pandas
>=2.2.2 -
scikit-learn
>=1.6 -
statsmodels
>=0.14.2 -
lifelines
>=0.27 -
tqdm
>=4.66 -
psutil
>=5.9 -
seaborn
>=0.13 -
formulaic
>=1.0
All dependencies are installed automatically when you install PyDTS.
Quick Start
The following example demonstrates how to generate synthetic data and fit a TwoStagesFitter model.
Detailed definitions and explanations are available in the methods section of the documentation.
The function generate_quick_start_df simulates a dataset with the following defaults:
- Sample size:
n_patients=10000 - Covariates:
n_cov=5independent covariates, each drawn from Uniform(0,1) distribution - Competing events:
j_events=2event types - Time scale:
d_times=14discrete time intervals - Hazard coefficients (default values):
- $\alpha_{1t}$ = −1 − 0.3 * log(t)
- $\alpha_{2t}$ = −1.75 − 0.15 * log(t)
- $\beta_1$ = −log([0.8, 3, 3, 2.5, 2])
- $\beta_2$ = −log([1, 3, 4, 3, 2])
For each patient, a censoring time $C$ is drawn from Uniform{1, ..., 14}. The observed time is defined as $X = min(T, C)$, where $T$ is the event time which is sampled based on the covariates of each patient and the hazard coefficients. If censoring occurs before the event ($C < T$), the event type is set to $J = 0$.
Once the dataset is generated, you can fit a TwoStagesFitter to the data (without columns $C$ and $T$ which are not observed in practice).
You can generate synthetic data and fit your first TwoStagesFitter model with the following code:
from pydts.fitters import TwoStagesFitter
from pydts.examples_utils.generate_simulations_data import generate_quick_start_df
# Generate a synthetic dataset with 10,000 patients,
# 5 covariates, 14 discrete time intervals, and 2 competing events
patients_df = generate_quick_start_df(n_patients=10000, n_cov=5, d_times=14, j_events=2, pid_col='pid', seed=0)
# Initialize and fit the discrete-time competing-risk model
fitter = TwoStagesFitter()
fitter.fit(df=patients_df.drop(['C', 'T'], axis=1))
# Display model summary
fitter.print_summary()
Citations
If you found PyDTS software useful to your research, please cite the papers:
@article{Meir_PyDTS_2022,
author = {Meir, Tomer and Gutman, Rom, and Gorfine, Malka},
doi = {10.48550/arXiv.2204.05731},
title = {{PyDTS: A Python Package for Discrete Time Survival Analysis with Competing Risks}},
url = {https://arxiv.org/abs/2204.05731},
year = {2022}
}
@article{Meir_Gorfine_DTSP_2025,
author = {Meir, Tomer and Gorfine, Malka},
doi = {10.1093/biomtc/ujaf040},
title = {{Discrete-Time Competing-Risks Regression with or without Penalization}},
year = {2025},
journal = {Biometrics},
volume = {81},
number = {2},
url = {https://academic.oup.com/biometrics/article/81/2/ujaf040/8120014},
}
and please consider starring the project on GitHub
How to Contribute
- Open Github issues to suggest new features or to report bugs\errors.
- Provide feedback on the documentation.
- Contact Tomer or Rom if you want to add a usage example to the documentation.
- If you want to become a developer (thank you, we appreciate it!) - please contact Tomer or Rom for developers' on-boarding.
Tomer Meir: tomer1812@gmail.com, Rom Gutman: rom.gutman1@gmail.com
Running Tests Locally
To run the test suite on your local machine, follow these steps:
- Clone the repository
git clone https://github.com/tomer1812/pydts.git
cd pydts
- Install the package in editable mode
pip install -e .
- Run the test suite with Poetry
pip install poetry
poetry run pytest
Release files for pydts 0.10.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| pydts-0.10.3.tar.gz | 1.4 MB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| pydts-0.10.3-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 2.8 MB
Release files / pydts-0.10.3.tar.gz
| Download URL | pydts-0.10.3.tar.gz |
|---|---|
| Size | 1.4 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8646e4bc13b8086981abcbd8b6e75e3cd2f46a755e7531e93ec725ed70f347d9
|
|
BLAKE2b-256 checksum How to use checksums |
b59fa825214be6e8fbb89a67f3a08f41cefa0316e6b385f4e78fa9d5d406aa44
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/1.5.1 CPython/3.9.19 Darwin/24.6.0
|
Release files / pydts-0.10.3-py3-none-any.whl
| Download URL | pydts-0.10.3-py3-none-any.whl |
|---|---|
| Size | 1.5 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
b1e72dc8221ddb5a3011369d80b8ec0b382f9f98f4cb547478ef53f155f099d9
|
|
BLAKE2b-256 checksum How to use checksums |
bb871dc5678838c85fd5872fb643dae29ccca5c96f876580c2d5701aec037946
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/1.5.1 CPython/3.9.19 Darwin/24.6.0
|
Security for pydts 0.10.3
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
