Anonymising a transcript is the removal of data that names a patient from a recorded session. When the visit covers substance misuse, the common-law duty of confidence and UK GDPR Art. 9 apply. anonym.plus runs on your device, so the dialogue stays while names are hidden.
When this applies
A clinic keeps visit transcripts for quality review. Each one names the patient, a relative, and a town, which must be stripped first.
How anonym.plus handles it
- Open the transcript in anonym.plus on your device.
- Local OCR reads any scanned or imaged text.
- It flags patient and relative names plus places.
- Review each flag and keep the dialogue intact.
- Replace each value with a label, or remove it.
- Save the clean transcript. The source stays local.
What you need to provide
- The transcript (PDF, DOCX, TXT, or scan).
- An operator: Replace, Redact, or Mask.
- Optional: a key map for re-linking.
Patient data entity types detected
| Category | anonym.plus entity type | Example |
|---|---|---|
| Patient | PERSON | Caleb Stone → [PATIENT_1] |
| Relative | PERSON | brother Theo → [FAMILY_1] |
| Town | LOCATION | Fairview → [PLACE] |
| Call date | DATE_TIME | Call 21/04/2026 → [DATE] |
| Phone | PHONE_NUMBER | +44 7700 900149 → [PHONE] |
| NHS number | NHS_NUMBER | NHS 812 040 1234 → [NHS_NO] |
Compliance achieved
- Removes identifiers in substance misuse dialogue under the common-law duty of confidence & UK GDPR Art. 9.
- On-device only, so no cloud processor contract is needed.
- Transcripts are sealed with AES-256-GCM.
- Assessed per ICO Anonymisation Code and UK GDPR Recital 26.
Anonymise telepsychiatry transcripts offline — see plans & start free →
Limitations & cautions
Spoken text drops names in mid-sentence and uses nicknames. The tool flags what it detects. You still scan the dialogue for an offhand name or place the model might miss.
Frequently asked questions
Why are transcripts tricky?
Speech is loose. People use nicknames and name a place in passing. The tool flags what it finds, but a careful human read still helps catch the rest.
Are relatives and places hidden?
Yes. A named relative or town is an identifier. The tool flags each so the patient cannot be traced through family or location.
Is anything uploaded?
No. The app runs on your own device. The transcript stays with you.