Health data · Research and teaching
Anonymize medical records while keeping every clinical detail
A discharge report that leaves the care pathway usually has a specific destination: a retrospective study, a teaching session with junior doctors, a case presented at a conference or a second opinion from a specialist at another hospital. In each case the diagnosis, progress and treatment are essential; the patient's name, record number and home address are not.
In Spain, plans are arranged with our sales team.
Health data is a special category under the GDPR, and identifiers turn up in unexpected places: the header, a scanned patient label, tables of vital signs and the free text of the clinical history. Anonimatum combines its own AI, running on EU servers, with custom patterns built around the format of your hospital's record numbers and health card numbers.
When a study needs to follow the same patient across several documents, blacking out is not enough. The «Healthcare» and «Research» profiles replace each identifier with a consistent substitute, shift dates while keeping the intervals between them, and produce an equivalence map held by whoever you decide, kept apart from the document the research team receives.
How to prepare a clinical report to leave the hospital
- 1
Define your identifiers
Create a custom pattern for your medical record number format and health card number, and save it for the whole department.
- 2
Choose: redact or pseudonymize
For teaching, anonymize with GDPR anonymization, Dates and Addresses. For a longitudinal study, use the «Research» profile, which keeps the gaps between dates.
- 3
Check the free text
In the preview, look closely at the history and progress notes, where relatives, clinicians or workplaces often appear.
- 4
Hand over the report, keep the map apart
The researcher gets the document; the equivalence map, in JSON or encrypted, stays with whoever is authorised to reverse it.
Fictitious document
Before and after: a pseudonymized discharge report
Fictitious discharge report from an invented hospital, processed with the «Healthcare» profile: the patient gets a consistent pseudonym and dates are shifted.
VALLEY DISTRICT HOSPITAL · Cardiology Department
Patient: Manuel Ortega Ruiz · Record no. 0047731 · Health card BBBB000000000000
Admitted: 03/02/2026 · Discharged: 09/02/2026
Address: Avenida de los Pinos 22, Villanueva · Phone: 600 000 000
Reason for admission: crushing chest pain lasting 40 minutes.
Diagnosis: non-ST-elevation acute coronary syndrome. Angiography shows 80% stenosis of the left anterior descending artery.
Discharge medication: aspirin 100 mg, ticagrelor 90 mg twice daily, atorvastatin 80 mg.
Signed: Dr Elena Campos Ibáñez, cardiologist.
VALLEY DISTRICT HOSPITAL · Cardiology Department
Patient: Jorge Salinas Prieto · Record no. 0081205 · Health card (redacted)
Admitted: 17/03/2026 · Discharged: 23/03/2026
Address: (redacted) · Phone: (redacted)
Reason for admission: crushing chest pain lasting 40 minutes.
Diagnosis: non-ST-elevation acute coronary syndrome. Angiography shows 80% stenosis of the left anterior descending artery.
Discharge medication: aspirin 100 mg, ticagrelor 90 mg twice daily, atorvastatin 80 mg.
Signed: Dr Eva Calvo Íñiguez, cardiologist.
The stay still lasts six days and the clinical content is untouched; only identifiers change, and the map can lead back to the real patient if your protocol allows it.
Built for the way hospitals actually write
Your hospital's own identifiers
Record and episode numbers follow no universal format. A custom pattern matching your format catches them in headers, tables and free text.
Dates that still work for analysis
The «Research» profile shifts dates while keeping the intervals between them, and keeps the ratios between amounts: length of stay or time to readmission can still be measured.
Images inside the report
Dermatology photos or screenshots with embedded identifiers are handled with the «Images and logos» option or with a manually drawn zone.
Spreadsheet extracts by column
If the study starts from a spreadsheet export, set the treatment per column: redact the name, pseudonymize the record number, leave clinical variables as they are.
FAQ: anonymizing medical records
Is a pseudonymized report still personal data?
Yes. As long as the equivalence map exists, the report is pseudonymized personal data and the GDPR still applies. Only truly anonymized data falls outside its scope.
Can I use it to send a case for a second opinion?
Yes. For a second opinion you remove what identifies the patient and keep tests, progress and treatment. If the report later has to go back into the patient's record, the «Healthcare» profile and its map let you reverse it.
What about scanned reports with patient labels?
OCR reads the label text before detection. If the label has a barcode or QR code, you can also cover it with a visual zone.
Who decides whether anonymization is enough?
Your organisation, based on the project's legal basis and the ethics committee's opinion. Anonimatum gives you both technical options, redaction or pseudonymization with a map, to apply that decision.
Do the reports leave the European Union?
No. Detection, OCR and anonymization run on EU infrastructure with Anonimatum's own models, without external large language model APIs. Documents are deleted after processing.
Sharing clinical reports for research or teaching?
Tell us which reports and formats you use and we will prepare a demo with fictitious documents that match yours.
Try it in the appNot in Spain?