BoXHED2.0: Scalable boosting of dynamic survival analysis

TL;DR

BoXHED2.0 is a scalable, nonparametric tree-boosted hazard estimator for complex survival analysis involving time-dependent covariates, supporting various survival settings with high computational efficiency.

Contribution

It introduces BoXHED2.0, a novel, scalable, nonparametric boosting method for diverse survival analysis scenarios, including time-dependent covariates, recurring events, and competing risks.

Findings

Achieves speed comparable to parametric models.

Supports complex survival settings beyond right-censoring.

Available as an open-source Python package.

Abstract

Modern applications of survival analysis increasingly involve time-dependent covariates. The Python package BoXHED2.0 is a tree-boosted hazard estimator that is fully nonparametric, and is applicable to survival settings far more general than right-censoring, including recurring events and competing risks. BoXHED2.0 is also scalable to the point of being on the same order of speed as parametric boosted survival models, in part because its core is written in C++ and it also supports the use of GPUs and multicore CPUs. BoXHED2.0 is available from PyPI and also from www.github.com/BoXHED.