AI Science Tracker
EN

A clinical de-identification model lost accuracy on intensive-care notes

Sharif University of Technology; Mayo Clinic; Yale University · 24 Oct 2026 · medicine

Discovery

A clinical de-identification model lost accuracy on intensive-care notes

Source check: Source checked. Not independent scientific confirmation.

Finding: EdgeDeID, trained on synthetic notes, scored 0.478 F1 on 2,434 MIMIC intensive-care nursing notes, compared with 0.994 on 100 synthetic notes. A separately trained clinical text model scored 0.675 on the intensive-care set. The test corpora differ in writing style and annotation conventions.

Technology: EdgeDeID; Qwen3-0.6B; clinical entity recognition

Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
Fine-tuned Qwen3-0.6B detects surrogate identifiers in clinical notes.
Test and result
External notes use replaced identifiers and human labels. The paper also reports an analysis limited to the 823 Safe Harbor spans.
Limitation / still unknown
The evaluation does not establish safe redaction or portability. Style and label differences both affect the synthetic-to-clinical comparison.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/abs/2610.33839v1 · https://arxiv.org/html/2610.33839v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 21 Nov 2026. Review scope: v1 §4.1 and Table 4 external clinical-text evaluation.

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse medicine discoveries.