SurvHive is a Python package that provides an interface to various survival-analysis models, from Cox to deep-learning-based ones, making it easier to access and compare different methods.
It is designed so that all method adapters are compliant scikit-learn estimators, allowing interoperability with scikit-learn facilities and it offers a range of features for model selection, parameter tuning, and performance evaluation.
From scikit-survival:
From Pycox:
From Auton Survival:
From LassoNet:
From SurvTRACE:
SurvHive provides multiple metrics for models evaluation, such as Harrell C-index, Antolini score, Brier score, and AUROC, and allows for the creation of user-defined ones.
See here for instruction.
See our guide.
See doc/ folder
Submitted for publication
This code is MIT-licensed (see included LICENSE file). The wrapped codes are licensed according to their own terms (mostly MIT).