How we use your data

Eka Care builds the models that transcribe your consultations, draft your notes, and help you extract discrete data from Medical Records. Making them better means learning from real clinical work. Our default policy is opt-in: we use your content unless you tell us not to. This page explains what that covers, what it never covers, and how to change it in one click.

1. THE SETTING
   1. Under Settings → Data & Privacy → Privacy, there is one control, titled Improve Eka Care's AI models. Its text reads: "We use your content, including but not limited to inputs, uploads, audio, transcripts, edits and generated outputs, to train, fine-tune and improve our AI models, unless you explicitly opt out through your account settings."
   2. Opting out is one click.
   3. This setting is separate from Data Retention, which governs how long We keep Your data whether or not model improvement is on.
2. WHO SEES YOUR DATA
   1. No person at Eka Care ever listens to a recording or reads a note. There is no review team, no transcription desk and no quality-assurance pool of humans reading consultations. The pipeline is automated from recording to model. Where an employee needs access to raw clinical data to resolve a fault, that access requires a support ticket, a time limit and Your clinic administrator's approval, and it is logged where You can see it.
   2. Identifiers are stripped before use, and audio is fragmented so a voice cannot be reassembled or matched to a person.
3. WHAT IS INCLUDED
   1. Included unless You opt out: consultation audio captured through EkaScribe, in clinic or on a teleconsultation; transcripts and the clinical notes generated from them; and everything You enter in the EMR, from diagnoses and prescriptions to observations, vitals and the free text You type Yourself.
4. WHAT WE DO BEFORE ANYTHING IS USED
   1. Text is stripped against the eighteen identifier categories of the HIPAA Safe Harbor method, covering names, all geography finer than a state, all dates except year, contact details, and record and device numbers. There is no Indian standard for this, so We use a published international one rather than inventing our own.
   2. Audio is segmented and shuffled so that the original recording cannot be reassembled. No voice signature is derived or stored.
   3. Re-identification testing is run on every dataset before it is used, and the result is recorded.
   4. The de-identified result is what our models see. Reversing it is not something We are able to do on request.
5. NOTHING LEAVES EKA CARE
   1. This setting is about improving Our own models. It does not permit Your data to go anywhere outside Eka Care, whether into a published research paper, a dataset that others can download, or a transfer to a research partner or a university.
   2. That is a separate decision, and it is switched off for every clinic on every plan. It cannot be enabled by a clinician from a settings page, and We cannot enable it on Your behalf. It requires the clinic’s own written decision, approval from an ethics committee, and a stated purpose for that specific piece of work.
6. HOW LONG WE KEEP THINGS
   1. The periods below apply whether or not model improvement is on. They are maximums, and a clinic may set anything shorter.

   Retention periods. Draft, pending confirmation by engineering and legal.

   What	Kept for	Why that long
   Original consultation audio	30 days	Long enough to redraft a note or resolve a dispute
   Transcripts	90 days	Clinical review and correction
   Clinical notes and records	Life of the account	They are the medical record; the clinic decides
   Consent records	7 years	So We can show what You agreed to, and when
   Processing logs	1 year after erasure	Required for audit
   Security incident logs	7 years	Security investigations
   De-identified training data	Until withdrawal	Not used once You opt out

7. WHERE YOUR DATA IS STORED
   1. All personal data covered by this page is stored and processed in India, on infrastructure We control. This is a product commitment and not a paid tier.
   2. Where a model from another provider is used to deliver a feature, We contract for zero data retention: the provider processes the request and keeps nothing, and is contractually barred from training on it. Where such terms are not available for a capability, We do not ship that capability.
8. CHANGING YOUR SETTING
   1. Open Settings, then Data & Privacy, then Privacy.
   2. Switch Improve Eka Care's AI models on or off. Opting out applies on the click. Opting back in asks You to confirm once.
   3. Your choice applies to every Eka Care surface You use, being the EMR, EkaScribe on web, the Chrome extension, the desktop application and the developer portal. It does not apply only to the surface on which You changed it.
   4. A patient wanting their own data removed from everything, and not only from model improvement, may use the QR code printed on their prescription or ask at the clinic. That request reaches the same erasure route and We confirm when it completes.
9. GRIEVANCE REDRESSAL
   1. If anything on this page does not match what You see in the product, please tell Us, and We will correct the product.
   2. Grievances relating to this page or to Your data may be addressed to the Grievance Officer named below, who will respond within the period prescribed under applicable law.

© 2026 Orbi Health Private Limited. This page forms part of the Eka Care Privacy Policy.