Patients and care team
The patient's name, but also those of physicians, department heads and referring doctors, wherever they appear.
Healthcare
De-identify your patient data on your own servers
Discharge summaries, referral letters, test results:
NymData detects and redacts the names, dates, addresses and contact details that identify the patient and the care team in your PDFs.
Medical content and layout are kept.
Your data never leaves your servers.
Every run is documented in an audit report.
Before: patient and physicians identifiable, in red
Department of Cardiology · Dr. Emily Brooks, Chief of Cardiology
Discharge Summary
Patient: John Carter, DOB 05/02/1981
Phone: (555) 013-2847
Address: 1428 Elm Street, Springfield, IL 62704
Patient admitted from 03/12/2026 to 03/15/2026 for exertional chest pain. Coronary angiography showed no significant lesion. Discharged on a beta blocker, follow-up in three months with Dr. Michael Reed.
Identities in the header and in the free text
After: identities redacted, medical content kept
Department of Cardiology · Dr. , Chief of Cardiology
Discharge Summary
Patient: , DOB
Phone:
Address:
Patient admitted from to for exertional chest pain. Coronary angiography showed no significant lesion. Discharged on a beta blocker, follow-up in three months with Dr. .
Same layout, identities removed from the file
- Patients and care team redacted
- Medical content kept
- PDF layout kept
- No data leaves your organization
How do you redact medical records in PDF?
You need to find everything that identifies the patient and the care team: names, identifiers, dates, addresses, contact details.
This protected health information (PHI) sits in the free text as well as in headers and footers.
With NymData, installed inside your organization, you select a folder of PDF clinical reports.
The detected data is removed from the file.
Medical content and layout are kept, and an audit report documents the run.
Nothing is sent outside.
The real challenge
Structured data is often easier to de-identify. Free text is still the hard part.
In a clinical data warehouse, structured data is often already pseudonymized.
Clinical reports, on the other hand, still contain identities in plain text.
They appear in the narrative written by clinicians as well as in each department's templates.
Free text and templates
History, conclusion, discharge letter.
But also template headers and footers, which change from one department to the next.
Patient identifiers
Names, dates and contact details are detected.
Why on your own servers
Sensitive data should not have to travel to be protected
Health data is a special category of personal data under the GDPR, and patient information is protected by medical confidentiality.
Sending it to an online service to de-identify it creates the very transfer you want to avoid.
Nothing leaves
Detection and redaction run on your servers.
No data or telemetry is sent out.
Controlled access
Administrator, operator, auditor: each sees what their role allows.
The auditor does not see document contents.
Traceability
Every run produces a timestamped, sealed audit report, ready for your records or your privacy officer.
Documents covered
What healthcare organizations de-identify
Everyday formats, processed in the same project.
- Clinical reports and letters in PDF, including those generated from Word: admissions, consultations, test reports. PDF · Word
- Patient record files in Excel or CSV: encounter lists, results, follow-ups. Excel and CSV
- Databases in PostgreSQL or MySQL feeding a data warehouse or a test database. Databases
- Presentations of clinical cases for teaching or conferences.
Use cases
When should you de-identify patient data?
Data warehouses and clinical research
Remove identities from clinical reports before using them in a clinical data warehouse or a research project.
As permitted by the approvals that govern your project.
Vendors and software providers
Share documents or a database with a vendor without handing over patient identities.
Testing clinical software
Populate test databases with realistic data, without real patients.
Teaching and AI
Use real cases for training or to train a model, without exposing patients.
Good to know
Anonymize or pseudonymize: what the GDPR says
Pseudonymizing (replacing a name with a constant identifier, for example) lets you follow the same patient across the whole project.
The data then remains personal health data, subject to the GDPR.
Effective anonymization takes data out of the scope of the GDPR, provided no re-identification is possible by reasonable means.
In healthcare, the medical information itself can be enough to re-identify someone: a rare disease, a small town, a precise date.
Check this risk before releasing a dataset.
NymData processes the data that identifies the person and documents every run.
Choosing the method and assessing the risk is done with your privacy officer.
Which method should you choose? · Pseudonymization vs anonymization: what is the difference?
Scope
What NymData handles in healthcare
- SupportedPDF clinical reports
- Including those generated from Word.
- Identifying data is removed from the file, the layout is kept.
- Word, Excel, CSV and PostgreSQL or MySQL databases are also supported.
- Good to knowLarge volumes
- For high volumes, the GPU option installs on your own hardware.
- Scheduled processing picks up the new documents in a folder every day.
- SupportedOn your infrastructure
- Installed on your organization's servers: nothing is sent outside.
- SupportedTraceability
- Every run produces a timestamped, sealed audit report for your privacy officer.
FAQ
Frequently asked questions about de-identifying patient data
Is patient data sent outside to be de-identified?
No. NymData is installed on your organization's servers.
Detection and redaction happen on your premises.
No data is sent to NymData or to any cloud service.
Is the medical information kept?
Yes. NymData removes the data that identifies the patient (name, date of birth, address, contact details).
The medical content stays readable, for research, training or testing.
Is pseudonymized data still health data under the GDPR?
Yes. Pseudonymized data is still personal data, and therefore health data subject to the GDPR.
Only effective anonymization, which prevents any re-identification by reasonable means, takes data out of the scope of the GDPR.
Are physicians' names redacted too?
Yes. Person names are detected throughout the document, whether they belong to the patient, the physicians or referring doctors.
This applies to free text as well as headers and footers.
Does the same patient keep the same identifier from one document to the next?
Yes, with the constant identifier method, in Word documents, Excel or CSV files and databases.
The same value always gets the same code within a project, so you can follow a patient from one document to another.
Can new clinical reports be de-identified automatically?
Yes. A project can be scheduled daily, weekly or monthly.
Only new or modified files are processed on each run.
See NymData redact PDF clinical reports, live
A demo on sample medical documents, remote or on site.
No patient data needed.