Monod, Melodie, Krusche, Peter, Qian, Cao, Berkman, Sahiner, Nicholas A., Petrick, Ohlssen, David and Coroller, Thibaud (2024) torchsurv: a lightweight package for deep survival analysis. Journal of Open Source Software.

Abstract

torchsurv is a Python package that serves as a companion tool for survival modeling
within the PyTorch framework. It offers functionalities for computing log-likelihoods of
common survival models and evaluating their predictive performance. torchsurv distin-
guishes itself by providing a simple and accessible programming interface with PyTorch
backend. Its lightweight design, requiring minimal input specifications and avoiding re-
strictive survival modelparameterizations, allows efficient model implementation for high-
dimensional input data. torchsurv is designed to support users, not to make them jump
through hoops by providing a simple yet rich set of survival tools.

Item Type:	Article
Keywords:	survival analysis; python; deep learning
Date Deposited:	04 Feb 2025 08:42
Last Modified:	04 Feb 2025 08:42
URI:	https://oak.novartis.com/id/eprint/53645