MedDeID enables hospitals to remove personal data from clinical notes locally
MedDeID enables locally governed clinical-text de-identification from real or synthetic training data
Computation and LanguageMachine Learning
Summary
Clinical notes often include personal information that makes it hard to use them for research without risking privacy. The authors created MedDeID, a tool that hospitals can run on their own computers to automatically find and hide personal details in clinical texts. The tool can learn from real or computer-generated notes, works well on Dutch and English data, and keeps sensitive information safe without sending it outside the hospital. This helps hospitals share valuable medical data for AI and research while protecting patient privacy.
What this means in practice
- •For hospital data teams: Automatically remove personal information from clinical notes on-site using MedDeID to enable safer data reuse for research and AI development.
- •For healthcare it providers: Integrate MedDeID into clinical software to offer privacy-protecting de-identification services that can run without sending data offsite.$Commercial implications: Allows creation of privacy tools for hospitals that comply with data residency laws and protect patient confidentiality.
Authors
Stig Hellemans, Tom Stroobants, Elyne Scheurwegs, Pieter Meysman, Philippe G. Jorens, Kris Laukens
Abstract
Clinical notes contain personally identifiable information (PII), restricting reuse for research and medical AI, especially when data cannot leave an institution. We developed MedDeID, an on-premises framework combining in-house annotation and synthetic-note generation with model training, inference, pseudonymisation and evaluation. On an independently annotated, adjudicated 300-note Dutch hospital benchmark, a hospital-trained compact transformer detected 98.9% of identifying text while redacting 0.24% of text outside annotated identifiers; a synthetic-only counterpart detected 96.1%. On 100 primary-care notes, the synthetic-trained model achieved higher recall than the hospital-trained model (90.3% versus 87.0%) and greater robustness to identifier-format perturbations. An English instantiation trained without real text detected 99.7% and 98.9% of annotated identifier characters on two external synthetic benchmarks. These results demonstrate transfer of the workflow to another language, but not clinical English performance. MedDeID provides a route to locally governed de-identification using real or synthetic training data.