The release of scikit-survival 0.24.0 introduces the cumulative_incidence_competing_risks() function, enhancing capabilities for survival analysis involving competing risks. This non-parametric estimator allows for more accurate modeling of events that can preclude each other. Additionally, compatibility with scikit-learn 1.6 is included, improving usability by supporting missing values in ExtraSurvivalTrees. The article also discusses how competing risks complicate standard survival analysis with real-world examples from a bone marrow transplant study involving leukemia patients, intimately connecting statistical modeling with practical healthcare scenarios.
The addition of cumulative_incidence_competing_risks() enhances scikit-survival, providing a non-parametric estimator for cumulative incidence functions amid competing risks.
Competing risks occur when multiple mutually exclusive events can prevent the occurrence of others, complicating traditional survival analysis.
This release includes support for scikit-learn 1.6, improving integration with additional tools and functionalities, including handling of missing values for ExtraSurvivalTrees.
The bone marrow transplant dataset exemplifies competing risks, analyzing 35 patients with different leukemia types and their outcomes post-transplant.
Collection
[
|
...
]