COPE: Chain-of-Thought Prediction Engine for open-source large language model-based stroke outcome prediction from clinical notes
Yongkai Liu, Helena Feng, Bin Jiang, Yixin Wang, Max Wintermark, David Liebeskind, Michael Moseley, Maarten Lansberg, Gregory Albers, Jeremy Heit, Greg Zaharchuk · 2026
Read COPE: Chain-of-Thought Prediction Engine for open-source large language model-based stroke outcome prediction from clinical notes by Yongkai Liu, Helena Feng, Bin Jiang, Yixin Wang, Max Wintermark, David Liebeskind, Michael Moseley, Maarten Lansberg, Gregory Albers, Jeremy Heit, Greg Zaharchuk (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/PMC13528327.txt.
Description
Yongkai Liu and ten colleagues build a reasoning language model to forecast recovery after stroke in a 2026 study. The tool, named COPE, reads discharge notes rather than neat data tables. Records covered 464 patients with ninety-day function scores. Two open-source models work in sequence. The first drafts clinical reasoning. The second turns it into a score prediction.
COPE matched the commercial GPT-4.1 closely on every metric. It also matched a strong model fed hand-built variables. Simpler text models and a one-step version fell behind. Errors grew for older patients and those with clot removal or long notes. Results held steady across sex and age groups. The authors stress privacy: open models can run inside hospitals. No retraining or manual feature work was needed.
Catalog record
| CREATOR | Yongkai Liu, Helena Feng, Bin Jiang, Yixin Wang, Max Wintermark, David Liebeskind, Michael Moseley, Maarten Lansberg, Gregory Albers, Jeremy Heit, Greg Zaharchuk |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40236 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |