Utilizing a domain-specific large language model for LI-RADS v2018 categorization of free-text MRI reports: a feasibility study
Mario Matute-González, Anna Darnell, Marc Comas-Cufí, Javier Pazó, Alexandre Soler, Belén Saborido, Ezequiel Mauro, Juan Turnes, Alejandro Forner, María Reig, Jordi Rimola · 2024
Read Utilizing a domain-specific large language model for LI-RADS v2018 categorization of free-text MRI reports: a feasibility study by Mario Matute-González, Anna Darnell, Marc Comas-Cufí, Javier Pazó, Alexandre Soler, Belén Saborido, Ezequiel Mauro, Juan Turnes, Alejandro Forner, María Reig, Jordi Rimola (2024) 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-11584817.txt.
Description
Liver MRI reports are written in free text, yet research needs each finding sorted into a shared scale called LI-RADS. Mario Matute-González and co-authors built a language model tuned for that sorting in 2024 and tested it on 291 small liver observations. The model read the report wording and assigned each finding its category.
Against a panel's agreed answers, the model agreed at a moderate level and caught the clearest cancer signs with high accuracy. Used as a first reader that flags easy cases, it lifted agreement sharply and cut the workload by about forty-five percent. The authors call it a workable helper for turning messy reports into clean research data. They stress it aids curation rather than replacing doctors.
Catalog record
| CREATOR | Mario Matute-González, Anna Darnell, Marc Comas-Cufí, Javier Pazó, Alexandre Soler, Belén Saborido, Ezequiel Mauro, Juan Turnes, Alejandro Forner, María Reig, Jordi Rimola |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |