Healthcare and business process outsourcing
Short answer: In healthcare and outsourcing work, files usually may not leave a designated network. EnjoyPDF processes documents without going online and the current version has no automatic network activity, so you can de-identify scans in batches on an isolated network and deliver a processing record that contains no source text.
Suitable work
- Research projects de-identifying medical records and test reports before giving them to a data analysis team;
- Outsourcing vendors processing clients’ scanned forms and removing identifying details that should not be kept before delivery;
- Removing personal information from sample files before giving them to a software vendor for troubleshooting.
Deployment recommendations
- Install on managed computers and agree not to use cloud AI translation (use a local model if needed). Silent MSI installation and central policy management are still planned.
- Download language packs in an environment with internet access, verify them, then import them.
- Block the software’s outbound connections with a firewall, and check this yourself once using the method in Verify that the software does not upload documents.
Processing workflow
See Document de-identification for research and outsourcing. Built-in rules cover ID numbers, bank card numbers, phone numbers and email addresses. Add names, medical record numbers, subject IDs and similar items as custom keywords.
Note
This page is not legal or compliance advice. Whether specific requirements of HIPAA, PIPL, PDPA or other regulations are met should be decided by your compliance officer.
FAQ
How do I install a language pack on an isolated network?
Download the language pack on a computer with internet access, copy it into the isolated network, and use "Import from file" in "Recognition language packs…". The SHA-256 is checked on import.
Will the software go online by itself?
Document processing does not go online. The current version connects only when you click to download a language pack or use cloud AI translation. In this kind of environment you can skip cloud AI translation (or use a local model) and import language packs from file.