Industries
Keep patient identifiers out of your AI prompts.
For practice admin and billing staff using an AI tool their organisation already approved. Identifiers are swapped on your device first.
The task
Billing staff spend hours on appeal letters for denied claims. The notes hold the patient's name, date of birth, record number and Medicare number. Many practices now draft these letters with an AI tool the organisation has approved.
Hushbridge works before the prompt is sent. On your device it swaps each identifier for a placeholder such as [MRN_1]. You check the text, then send it. The real values return on your device when the draft comes back.
Read this before you use it with patient information
Hushbridge is not a HIPAA business associate. Hushbridge signs no BAA. Its output is not de-identified under HIPAA. Hushbridge reduces what reaches the AI, but it does not make an AI tool approved for patient records. Follow your organisation's policy and ask your privacy officer before you use any AI tool with patient information.
What the healthcare profile adds
The healthcare profile protects every full date, not only dates of birth. Admission, discharge and visit dates are swapped too. It recognises NPI numbers by their Luhn check, DEA numbers by their check digit and Medicare MBIs by their format. Record numbers and health plan IDs are caught when they follow their labels.
Before and after
Fictional example
What you paste
Draft an appeal letter for patient Bartholomew Okafor, DOB 04/17/1958, MRN: 00482913, Medicare MBI 1EG4-TE5-MK73. He was seen on 08/02/2026 by Dr. Imelda Sorensen, NPI 1234567893. The claim for that visit was denied as not medically necessary.
What the AI gets
Draft an appeal letter for patient [PERSON_1], DOB [DOB_1], MRN: [MRN_1], Medicare MBI [MBI_1]. He was seen on [DATE_1] by Dr. [PERSON_2], NPI [NPI_1]. The claim for that visit was denied as not medically necessary.
7 details swapped: 2 names, 1 date of birth, 1 record number, 1 Medicare MBI, 1 visit date, 1 NPI. When the reply comes back, the real values are restored on your device. Run with the healthcare profile. The NPI and MBI are published sample formats. All names, numbers and addresses are made up.
What you review before sending
Clinical detail can identify a patient even with every number swapped. Check these first.
- Diagnoses, procedures and rare conditions that point to one person.
- Partial dates such as "March 2026" or "last Tuesday". Only full dates are swapped.
- Ages under 90. Only ages of 90 and over are swapped.
- Facility, ward and town names, and relatives named in the notes.
- ID numbers in your own system format with no label next to them.
- Scanned forms, images and PDFs.
Limits
- Detection uses rules, checksums and name lists. It can miss a name that has no title, role or label next to it. Clinical context can still identify a patient.
- The extension does not scan attachments. Files you upload to a chat site go as they are.
- A chat site's page can read text while you type. Prepare notes in the free check and paste only the protected version.
- Addresses are found in US style only. Names in non-Latin scripts are not covered.
- Review before sending stays necessary, and so does your organisation's approval of the AI tool.
Try the free check with made-up patient details first. It runs in your browser and sends no text to an AI model.
Try it with your own text