Use Case: Healthcare

Anonymize patient records for research, clinical trials, and external sharing — entirely offline.

Who This Is For

Patient data anonymization needs differ across a health system. anonym.plus is built for the people who have to move PHI without moving risk:

The Challenge

Challenge

A hospital research department needs to share patient outcome datasets with external academic partners for a multicenter clinical study. The datasets contain names, dates of admission, diagnoses, medication codes, and national patient IDs. Uploading to any cloud service risks GDPR Art. 9 special-category data violations and HIPAA PHI transmission requirements. No Business Associate Agreement has been established with the research partner's IT infrastructure.

Why Healthcare Is the Highest-Risk Industry for Breaches

Every one of the incidents below started the same way: patient data left a controlled environment. That is exactly the step anonym.plus is designed to remove from your workflow.

StatisticDetailSource
$7.42MAverage cost of a healthcare data breachHIPAA Journal, 2025
139 breachesLarge healthcare breaches reported in Q3 2025 alone, affecting 9.5 million patientsCompliancy Group, 2025
193 million peopleAffected by the Change Healthcare breach — the largest healthcare data breach on recordHHS OCR breach portal; HIPAA Journal, 2025
Costliest sectorHealthcare has repeatedly had the highest average data-breach cost of any industry in IBM's annual reportIBM Cost of a Data Breach Report

These figures describe breaches at organizations that, in most cases, already had cloud-based security tooling in place. Anonymizing PHI before it is uploaded anywhere — rather than trusting a vendor to protect it after upload — removes an entire category of exposure.

The Solution

Solution

The research team installs anonym.plus on a workstation within the hospital's secure network. They use Batch mode to process the full patient dataset (XLSX format, 40,000 rows) using the Healthcare preset. All 18 HIPAA Safe Harbor identifiers are removed using the Replace operator. Health data is permanently replaced with structured labels: [PERSON_1], [DATE_1], [MEDICAL_RECORD_1]. The processing takes under 3 minutes. No data leaves the hospital network during processing.

The Results

Result
  • All 18 HIPAA Safe Harbor identifiers removed — Safe Harbor de-identification achieved
  • Dataset exits GDPR Art. 9 scope — no longer health data under Recital 26
  • No BAA required — anonym.plus is not a Business Associate; no PHI transmitted to any third party
  • Dataset shared securely with academic partners via standard file transfer — no special data transfer agreement needed
  • Processing history logged locally as documentation for DPIA and research ethics compliance

PHI Entity Types Detected

HIPAA Categoryanonym.plus Entity TypeExample
NamesPERSONDr. Maria Schmidt → [PERSON_1]
DatesDATE_TIME03/15/1978 → [DATE_1]
Phone numbersPHONE_NUMBER+49 30 123456 → [PHONE_1]
Email addressesEMAIL_ADDRESSpatient@clinic.de → [EMAIL_1]
SSN / National IDsUS_SSN, DE_TAX_ID, IN_AADHAAR123-45-6789 → [ID_1]
Medical record numbersMEDICAL_LICENSE, Custom entityMRN-78234 → [MRN_1]
IP addressesIP_ADDRESS192.168.1.45 → [IP_1]
Geographic dataLOCATION, STREET_ADDRESSHauptstraße 12, Berlin → [LOCATION_1]

Compliance Achieved

How Offline De-Identification Works

  1. Load locally. Drag in clinical notes, spreadsheet exports, or EHR data extracts (CSV, XLSX, or plain text). Nothing uploads — the file only ever exists on the workstation running anonym.plus.
  2. Detect all 18 Safe Harbor identifiers. The local detection pipeline (Presidio + spaCy models running on-device) flags names, dates, MRNs, device identifiers, IP addresses, and the rest of anonym.plus's 340+ entity types — no data leaves the machine during scanning.
  3. Choose an operator and export. Use Replace, Redact, or Mask for permanent, irreversible de-identification, or Encrypt when a study may need to re-identify specific records later — Encrypt uses a locally stored AES-256-GCM key, decryptable only on a device that holds that key.

On-Device vs. Cloud vs. Manual Review

Dimensionanonym.plus (on-device)Cloud PII/DLP toolsManual chart review
PHI leaves your networkNever — processing is 100% localYes — uploaded to a vendor's servers or APINo upload, but no consistent audit trail either
BAA requiredNo — anonym.plus is not a Business Associate; no PHI is transmittedUsually, since PHI is processed by a third partyN/A
Throughput on large datasetsBatch mode processes tens of thousands of rows locally in minutesLimited by API rate limits and network conditionsHours to days, entirely human-paced
Entity coverage340+ entity types, including all 18 HIPAA Safe Harbor identifiersVaries significantly by vendor and planLimited to what the reviewer is trained to recognize
Ongoing costOne-time license, no subscriptionRecurring subscription or usage-based feesOngoing staff time
Documentation for complianceLocal processing log supports DPIA/HIPAA audit documentationDepends on vendor's logging/export capabilityManual notes, inconsistent

Why On-Device Wins for Healthcare Specifically

PHI is uniquely valuable and uniquely regulated. Every tool that requires uploading patient data to a third-party server or API introduces a new vendor relationship, a new potential Business Associate determination, and a new point where a breach can occur outside your control. anonym.plus runs the entire detection-and-de-identification pipeline on the same workstation that is already inside your facility's network perimeter — patient data is never transmitted anywhere during processing. You can verify this yourself by disconnecting the network entirely — anonym.plus still processes documents with zero outbound network calls, by design.

This matters most for the two situations healthcare organizations face repeatedly: sharing a dataset with an external research partner before a data use agreement is fully negotiated, and responding to an audit or records request under time pressure. In both cases, the fastest and lowest-risk path is to remove the identifiers before the data leaves the building — not to trust a new cloud processor to remove them after it arrives.

Ready to try? Read the HIPAA de-identification guide →

Important Considerations

Frequently Asked Questions

Does anonym.plus handle HIPAA-regulated patient records?

Yes. All 18 Safe Harbor PHI categories are covered by built-in entity types plus custom entity support. Processing is 100% offline. anonym.plus is not a Business Associate — no BAA required or available.

Does anonymizing with anonym.plus satisfy GDPR Article 9 for health data?

Yes. Once processed with Replace or Redact operators, health data is no longer personal data under GDPR Recital 26. Art. 9 special-category restrictions no longer apply to the anonymized output.

What file size can anonym.plus handle for large patient datasets?

XLSX: 20 MB / 100,000 rows. CSV: 30 MB. For larger datasets, split into batches. Batch mode processes up to 20 files in parallel (Pro plan).