Real-world evaluation of a transformer-based natural language processing system for identifying social determinants of health from routine clinical documentation
Xiangren Wang, Jingchuan Guo, Yi Guo, Yao An Lee, Mattia Prosperi, Jiang Bian, Noah Hammarlund · 2026
Read Real-world evaluation of a transformer-based natural language processing system for identifying social determinants of health from routine clinical documentation by Xiangren Wang, Jingchuan Guo, Yi Guo, Yao An Lee, Mattia Prosperi, Jiang Bian, Noah Hammarlund (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/PMC13427761.txt.
Description
Wang and colleagues evaluated in 2026 a transformer-based language system, called SODA, that reads clinical notes to extract social determinants of health such as housing, transport, and money strain. At a Florida health system, they compared the system's output against a research survey. The comparison covered 414 patients who also had notes and a standard electronic questionnaire available.
The system matched the survey best when patients reported no social needs. Sensitivity for real needs was weak in most of nine domains. It caught about fifty-five percent of alcohol use, but only five percent of abuse, sixteen percent of financial strain, and no drug use cases at all. The two surveys also agreed only modestly with each other. The authors conclude there is no perfect reference standard, and gaps reflect both what clinicians write down and what patients choose to disclose.
Catalog record
| CREATOR | Xiangren Wang, Jingchuan Guo, Yi Guo, Yao An Lee, Mattia Prosperi, Jiang Bian, Noah Hammarlund |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |