{"id":"arxiv/2610.19643","url":"https://arxiv.org/abs/2610.19643","tags":["stat.ME","math.ST"],"score":0,"title":"Model-Induced Hazard Geometry and Functional Representations in Survival Analysis","domain":"stat.ME","source":"arxiv","authors":["Xijia Liu"],"reviews":0,"subject":"stat.ME","summary":"Individualized survival predictions are often summarized by scalar risk scores or event probabilities at selected time horizons, although clinically relevant risk may also differ in its temporal organization. We consider finite-dimensional representations of individual hazard trajectories through interpretable hazard functionals, such as peak timing, cumulative hazard over prespecified intervals, and persistence of elevated risk. A central issue is that the richness of such representations is constrained by the hazard geometry supported by the underlying survival model. We formalize this…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-10-14","reactions":{}}