GRAFT: Decoupling Ranking and Calibration for Survival Analysis
Mohammad Ashhad, Robert Hoehndorf, Ricardo Henao · 2026
Read GRAFT: Decoupling Ranking and Calibration for Survival Analysis by Mohammad Ashhad, Robert Hoehndorf, Ricardo Henao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-07884.txt.
Description
Survival analysis must handle cut data, many traits, and curved effects, yet old models tune well but stay rigid, while deep models rank well but miscalibrate. Mohammad Ashhad and two co-authors propose GRAFT, a model that splits the job of ranking patients by risk from the job of tuning survival odds.
A linear fast-failure-time model pairs with a curved extra network, random gates select traits automatically, and training uses a ranking loss aligned with the match index plus local Kaplan-Meier fill-in. Tuned estimates follow from simple post-training calibration. On public benchmarks GRAFT beats baselines on both discrimination and calibration while staying sparse under noise, a combination clinicians and biostatisticians should welcome.
Catalog record
| CREATOR | Mohammad Ashhad, Robert Hoehndorf, Ricardo Henao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-07884