Search papers, labs, and topics across Lattice.
The paper introduces Sysurv, a differentiable, non-parametric method for identifying subgroups with exceptional survival characteristics without restrictive assumptions about the survival model. Sysurv leverages random survival forests to learn individual survival curves and uses these to automatically learn interpretable rules for subgroup selection. Experiments on various datasets, including cancer data, demonstrate Sysurv's ability to discover insightful and actionable survival subgroups, outperforming existing methods.
Ditch the restrictive assumptions: Sysurv uses differentiable random survival forests to automatically discover interpretable subgroups with exceptional survival rates.
In many applications, it is important to identify subpopulations that survive longer or shorter than the rest of the population. In medicine, for example, it allows determining which patients benefit from treatment, and in predictive maintenance, which components are more likely to fail. Existing methods for discovering subgroups with exceptional survival characteristics require restrictive assumptions about the survival model (e.g. proportional hazards), pre-discretized features, and, as they compare average statistics, tend to overlook individual deviations. In this paper, we propose Sysurv, a fully differentiable, non-parametric method that leverages random survival forests to learn individual survival curves, automatically learns conditions and how to combine these into inherently interpretable rules, so as to select subgroups with exceptional survival characteristics. Empirical evaluation on a wide range of datasets and settings, including a case study on cancer data, shows that Sysurv reveals insightful and actionable survival subgroups.
