pytorch
💡 [REQUEST] - Tutorial on deep survival analysis using PyTorch & TorchSurv
🚀 Describe the improvement or the new tutorial
TorchSurv is a Python package that serves as a companion tool to perform deep survival modeling within the PyTorch environment. Unlike existing libraries that impose specific parametric forms on users, TorchSurv enables the use of custom PyTorch-based deep survival models. With its lightweight design, minimal input requirements, full PyTorch backend, and freedom from restrictive survival model parameterizations, TorchSurv facilitates efficient survival model implementation, particularly beneficial for high-dimensional input data scenarios.
In this tutorial, we want to introduce how to easily use our package, from loss functions (Weibull and Cox model), evaluation metrics (concordance-index, AUC, Brier score) and statistical tools (Kaplan-Meier, estimator). This will enable Pytorch users to develop true survival model by changing few lines of code while using their favorite deep learning framework!
Existing tutorials on this topic
The tutorial will be adapted from our existing documentations:
Additional context
category: survival analysis
This work was made as part of the collaboration research between the FDA and Novartis
Further read:
- Our preprint manuscript can be found here.
- Features comparison between best
RandPythonpackages can be found in this section - Performance benchmarks and evaluations can be found here
Hello @svekars and @albanD, would you have any thoughts on the request? Our package has been downloaded over 1k last month alone (after only few months since its initial release). We truly believe that this tool will be used by academics as well as industries as there is a gap in reliable package for deep survival learning. We are committed to actively maintain and improve it, and being part of the PyTorch tutorial collection would allow us to reach new users and strengthen our package.
Thanks, Thibaud
Hi @tcoroller, there is a process to add ecosystem partners. Can you please fill out this form: https://pytorch.org/ecosystem/join and they will follow up with you. After the tool is added to the ecosystem page, we'll be happy to work with you on publishing the tutorials.