CURE-OOD: Benchmarking Out-of-Distribution Detection for Survival Prediction

2026-04-30

AI Summary

This study addresses the lack of out-of-distribution (OOD) detection capabilities in cancer survival prediction, where variations in imaging acquisition protocols lead to distributional shifts. To this end, the authors introduce CURE-OOD, the first benchmark specifically designed for OOD detection in survival analysis, which partitions CT scans into training, in-distribution, and out-of-distribution test sets based on imaging parameters. The work demonstrates that conventional classification-oriented OOD detection methods exhibit significant limitations in this context and proposes HazardDev, a survival-aware baseline tailored to the task. Experimental results show that covariate shift substantially degrades survival prediction performance, while CURE-OOD enables systematic evaluation of model robustness and OOD detectability in survival analysis.

Abstract

``How long can I live and remain free of cancer?'' is often the first question a patient asks after receiving a cancer diagnosis and treatment. Accurate survival prediction helps alleviate psychological distress and supports risk stratification and personalized treatment planning. Recent survival prediction frameworks have shown strong performance using computed tomography (CT) images. However, variations in imaging acquisition introduce out-of-distribution (OOD) samples caused by covariate shifts that undermine model reliability. Despite this challenge, to our knowledge, no existing benchmark systematically studies OOD detection in cancer survival prediction. To address this gap, we introduce the Cancer sURvival bEnchmark for OOD Detection (CURE-OOD), the first benchmark for systematically evaluating OOD detection in survival prediction under controlled acquisition-induced distribution shifts. CURE-OOD defines scanner-parameter-based training, in-distribution (ID), and OOD test splits across four survival prediction tasks. Our experiments show that covariate shifts notably reduce survival prediction performance. It also shows that mainstream classification-oriented OOD detectors can fail in survival prediction. Finally, we include HazardDev as a simple survival-aware reference baseline for OOD detection. CURE-OOD enables systematic analysis of how distribution shifts affect both downstream survival performance and OOD detectability.

Problem

• out-of-distribution detection
• survival prediction
• covariate shift
• benchmark
• cancer

Innovation

• out-of-distribution detection
• survival prediction
• distribution shift
• medical imaging
• benchmark