The risk posed by a tumor can be described on two axes: how far the tumor has already grown and spread in the body (staging) and how dangerous/(de-)differentiated the tumor is irrespective of staging due to its pathological features (grading). When we apply survival analysis to cancer data, we implicitly mix staging and grading. Our goal is to develop novel survival analysis approaches to support medical experts in deriving a grading scheme from data. Hence, we need models that are small and sparse enough for medical experts to analyze, take bio-medical constraints into account, are still predictive of survival, and, most importantly, are decorrelated from staging indicators. Learning such models from small data is a novel machine learning problem that we dub decorrelated sparse survival regression (DSSR). In this paper, we provide three algorithms for DSSR, develop further mechanisms for model interpretability, and evaluate all algorithms on both simulated and real-world cancer data. • We define the machine learning problem of decorrelated sparse survival regression (DSSR) with applications in tumor grading. • We provide three novel algorithms to address the problem. • We evaluate the algorithms on both simulated and real-world data and find substantial improvements beyond prior algorithms, especially in terms of interpretability.

Exception handlers—and effect handlers more generally—are language mechanisms for structured nonlocal control flow. A recent trend in language-design research has introduced lexically scoped handlers, which address a modularity problem with dynamic scoping. While dynamically scoped handlers allow zero-overhead implementations when no effects are raised, existing implementations of lexically scoped handlers require programs to pay a cost just for having handlers in the lexical context. In this paper, we present a novel approach to implementing lexically scoped handlers of exceptional effects. It satisfies the zero-overhead principle—a property otherwise met by most modern compilers supporting dynamically scoped exception handlers. The key idea is a type-directed translation that emits information indicating how handlers come into the lexical context. This information guides the runtime in walking the stack to locate the right handler. Crucially, no reified lexical identifiers of handlers are needed, and mainline code is not slowed down by the presence of handlers in the program text. We formalize the essential aspects of this compilation scheme and prove it correct. We integrate our approach into the Lexa language, allowing the compilation strategy to be customized for each declared effect based on its expected invocation rate. Empirical results suggest that the new Lexa compiler reduces run-time overhead in low-effect or no-effect scenarios while preserving competitive performance for effect-heavy workloads.