SigBERT: Combining Narrative Medical Reports and Rough Path Signature Theory for Survival Risk Estimation in Oncology
Paul MINCHELLA, Loïc VERLINGUE, Stéphane CHRETIEN, Rémi VAUCHER, Guillaume METZLER · 2025
Read SigBERT: Combining Narrative Medical Reports and Rough Path Signature Theory for Survival Risk Estimation in Oncology by Paul MINCHELLA, Loïc VERLINGUE, Stéphane CHRETIEN, Rémi VAUCHER, Guillaume METZLER (2025) 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-2507-22941.txt.
Description
Cancer death risk from doctors' notes is scored by SigBERT, a tool from Paul Minchella and co-authors in a 2025 arXiv paper. Patient files hold rich text that survival tools read poorly in time order. SigBERT boils each note to a sense vector, then reads the trail of vectors as a path. A rough-path mark, the signature, sums that path's shape into features.
Those features feed a lean survival model that scores each patient's risk. On a real French cancer center set, the score sorted outcomes well on held-out cases. Time shape in notes, it seems, carries real risk signal.
Catalog record
| CREATOR | Paul MINCHELLA, Loïc VERLINGUE, Stéphane CHRETIEN, Rémi VAUCHER, Guillaume METZLER |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-22941