A generative approach for semantic auditing of electronic health records
Irena Girshovitz, Atai Ambus, Moni Shahar, Ran Gilad-Bachrach · 2026
Read A generative approach for semantic auditing of electronic health records by Irena Girshovitz, Atai Ambus, Moni Shahar, Ran Gilad-Bachrach (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-13522542.txt.
Description
Irena Girshovitz and colleagues proposed testing health records the way engineers test software. Published in 2026, the method, called Medical Data Pecking, uses language models to write context-aware checks that peck at gaps between recorded data and known disease patterns. Current record checks catch syntax slips or need hand-built rules, missing deeper meaning errors that can mislead clinical AI.
A reference tool built on retrieval-augmented generation turned medical literature into runnable checks. Applied to three datasets, it produced dozens of tests per group and flagged mismatches between observed rates and expected patterns. Some flags marked true data errors, while others reflected how patient groups were chosen. The authors offer a first framework for scalable semantic auditing, shifting trust from manual rules to generative verification.
Catalog record
| CREATOR | Irena Girshovitz, Atai Ambus, Moni Shahar, Ran Gilad-Bachrach |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |