Extracting Cardiorespiratory Symptoms From Clinical Notes Using Open-Weight Large Language Models: Method Development and Validation Study
Yunbing Bai, Wanting Cui, Joseph Finkelstein, Andrew Coristine · 2026
Read Extracting Cardiorespiratory Symptoms From Clinical Notes Using Open-Weight Large Language Models: Method Development and Validation Study by Yunbing Bai, Wanting Cui, Joseph Finkelstein, Andrew Coristine (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/pmc-13216758.txt.
Description
Yunbing Bai and co-authors present a 2026 informatics study on pulling heart and lung symptoms out of free-text clinical notes automatically, then converting them into standard diagnosis codes. Symptom data are central to early detection, yet most of it sits trapped in narrative text.
The team tested locally run open-weight language models on 593 notes, keeping all data on site for safety. Four prompting strategies were compared, from bare instructions to a multimodule pipeline with post-processing; more structure consistently helped. On held-out notes, the gpt-oss-120B model clearly beat Llama 3.3-70B, reaching F1 scores of 0.88 for symptom extraction and 0.87 for code mapping. The authors argue such pipelines can scale symptom surveillance without exporting patient data.
Catalog record
| CREATOR | Yunbing Bai, Wanting Cui, Joseph Finkelstein, Andrew Coristine |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 54 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |