Automatic identification of diagnosis from hospital discharge letters via weakly supervised Natural Language Processing
Vittorio Torri, Elisa Barbieri, Anna Cantarutti, Carlo Giaquinto, Francesca Ieva · 2026
Read Automatic identification of diagnosis from hospital discharge letters via weakly supervised Natural Language Processing by Vittorio Torri, Elisa Barbieri, Anna Cantarutti, Carlo Giaquinto, Francesca Ieva (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-13507187.txt.
Description
Reading diagnoses out of discharge letters takes armies of annotators, and Vittorio Torri and co-authors dodge most of that labor with weak labels, in work from twenty twenty-six. Their pipeline pulls diagnosis-bearing sentences, embeds them with a model tuned on Italian medical text, and clusters them twice to mint labels. A document classifier then trains on the minted set.
On more than thirty thousand children's letters about a chest ill, the best model scored near fully taught rivals. Hand-label hours saved topped fifteen hundred. A second, smaller chest set ranked the models the same way. Big record studies can start without the annotation wall.
Catalog record
| CREATOR | Vittorio Torri, Elisa Barbieri, Anna Cantarutti, Carlo Giaquinto, Francesca Ieva |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 226 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |