Privacy and Data Handling — Sanavex
https://sanavex.si/privacy/
Plain-text version for low-bandwidth connections. Full page: https://sanavex.si/privacy/

Institutional Protocol

# Privacy and Data Handling
This page describes how Sanavex handles data during the pilot, as it is today. We will update it before any real patient document is used.

## No model training

Sanavex does not train AI models on patient data without explicit consent. To produce a summary, documents are processed by a third-party AI service.
Our commitment

## Security principles
-

### Encryption in transit
Documents travel over encrypted connections. They are processed on our servers and by an AI service to produce a summary, so they are not end-to-end encrypted.

-

### Access Controls
Access is limited by role.

## Information we collect

To produce a summary, we process the documents a doctor uploads or photographs: lab reports, prescriptions, discharge summaries and referral letters.
Uploaded documentsPhotographs or scans of printed medical documents.

MetadataTimestamps, user identifiers, and institutional routing tags.

## Pilot document handling

Phase 1 of the pilot uses simulated documents only. No real patient documents are uploaded. We will publish retention periods before that changes.

## Patient consent responsibility
Before any real patient document is uploaded, the clinic is responsible for informing the patient and recording consent. Sanavex does not contact patients directly.

Sanavex <https://sanavex.si/>ساناڤكس <https://sanavex.si/ar>
Sanavex develops clinician-supervised AI tools for Syrian healthcare workflows.
© 2026 Sanavex

### Product
- RecordFlow <https://sanavex.si/recordflow>
- Pilot Program <https://sanavex.si/pilot>
- Apply for Pilot <https://sanavex.si/pilot/apply>

### Company
- Approach <https://sanavex.si/approach>
- About <https://sanavex.si/about>
- Team <https://sanavex.si/team>
- Contact <https://sanavex.si/contact>

### Legal
- Privacy Policy <https://sanavex.si/privacy>

RecordFlow does not provide diagnosis, treatment recommendations, or emergency decision support. AI output is not valid for clinical use until reviewed and approved by a clinician.