Automated phenotyping of ophthalmologic diseases from routine medical records using small language models and the human phenotype ontology (HPO)
Binh Duong Thai, Sebastian Arens, Thomas Reinhard, Daniel Böhringer · 2026
Read Automated phenotyping of ophthalmologic diseases from routine medical records using small language models and the human phenotype ontology (HPO) by Binh Duong Thai, Sebastian Arens, Thomas Reinhard, Daniel Böhringer (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-13157479.txt.
Description
Binh Duong Thai, Sebastian Arens, Thomas Reinhard, and Daniel Böhringer published a 2026 study on teaching computers to extract standardized eye-disease terms from doctors' notes. Clinicians write findings in free text, which resists comparison and research. A shared catalog, the Human Phenotype Ontology, names traits precisely, but coding notes by hand is slow. The team built a local pipeline around a small language model.
The system splits text, spots negations, and matches phrases against an enriched multilingual catalog. It was trained on anonymized doctors' letters. For testing, 175 real record excerpts were coded by hand, yielding 342 catalog terms. The pipeline found 341 of them. Its overlap with the human coding was solid, with precision and recall both near 0.82. The authors argue such a tool could lighten coding work in eye clinics and make records easier to share and study.
Catalog record
| CREATOR | Binh Duong Thai, Sebastian Arens, Thomas Reinhard, Daniel Böhringer |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |