Turning the doctor's voice into a report in an OB/GYN clinic: consent, the exam, and accuracy in Arabic
You talk with your eyes on the patient, and the visit is written up on your sheet when she leaves. That's what voice notes in the clinic promise. In an OB/GYN clinic in particular, it's also what needs the most care: the patient tells you things she tells no one else, and she's exposed during the exam. We read what's been published so far on how patients feel, on consent, on how well Egyptian Arabic is understood, and on how much time it really saves, and collected what's useful for your clinic.
What worries the patient?
The largest published survey asked patients before a university hospital started using voice recording, and 1,893 of them answered (UC Davis, 2025). 48% welcomed it, 33% didn't mind, and 19% were against it. When they wrote what worried them, accuracy came first (39%), then privacy (15%), then being recorded at all (13%).
Another study in New York (JAMA Network Open, 2025) found two things that matter directly to an OB/GYN clinic:
- Patients hold things back. 41% said they'd talk less about sexual health with the recording on.
- They want to know where their voice goes. 96 to 98% said how the recording is used, where it's sent and who hears it mattered to their decision.
In May 2026, an OB/GYN in New York wrote about a new patient who agreed, after some hesitation, to recording during a routine gynaecological exam, then complained days later that she had "felt violated". The exam itself had gone well. What hurt was having something foreign present in an intimate moment. The doctor stopped recording in her own practice after that.
The lesson isn't that recording doesn't belong in an OB/GYN clinic. It's that consent needs more care here, and that the exam has its own rule.
Consent: before the visit, in one clear sentence, and easy to refuse
In Egypt, the Personal Data Protection Law (No. 151 of 2020) treats health data as sensitive data, and processing it requires the person's explicit written consent, except where the law allows otherwise. The Code of Medical Ethics (2003) forbids a doctor from disclosing a patient's secrets. In Saudi Arabia, the Personal Data Protection Law also treats health data as sensitive. We haven't found guidance from any health ministry or medical syndicate in the region on voice recording specifically, so the closest guide is what clinics that started earlier do, and what medical bodies in the UK and Canada say:
- She knows before she walks in. A line in the booking confirmation, or a notice in the waiting room: "The doctor may record the visit by voice to write your report. You can say no, and nothing about your visit will change."
- She's asked in the room, in a clear sentence, and every question gets an answer. In the New York study, consent fell from 82% to 55% once patients learned the details of storage and the company involved. Her questions about where her voice goes are real, and only a true answer will reassure her.
- Saying no is easy, and costs her nothing. No look, no comment. The UK's General Medical Council (GMC) says patients can withdraw consent during or after a recording without it affecting their care.
- Her consent is written in the visit. And since Egyptian law asks for written consent to process health data, ask your lawyer for a form she signs once when her file is opened.
A sentence you could say, in Egyptian Arabic: «أنا بسجّل الكلام عشان التقرير يتكتب وعيني عليكي مش على الورق. التسجيل بيتمسح بعد ما أراجع التقرير، وهوقفه وقت الكشف. لو مش مرتاحة، مفيش أي مشكلة.» ("I'm recording so the report writes itself while I look at you, not at the paper. The recording is deleted once I've checked the report, and I'll stop it during the exam. If you're not comfortable, that's completely fine.")
During the exam: stop the recording
We found no published guidance that explicitly says to stop recording during a gynaecological exam. The closest is that the GMC asks you to stop if the patient asks or if it's disrupting the consultation, and NHS England's guidance says recording may not suit highly sensitive conversations. Here's what we recommend for an OB/GYN clinic:
- Stop the recording before she moves to the examination couch, and say so out loud: "I've stopped the recording."
- Once she's dressed, dictate what you found on examination and ultrasound in a minute.
- If something comes up that she doesn't want written down (a past pregnancy, a relationship, violence), stop it, and write what's needed yourself.
- The rule for a chaperone during the exam stays the same. Recording doesn't replace one, and a chaperone doesn't replace consent.
Accuracy in Egyptian Arabic: a draft, not a report
A doctor in Egypt speaks Egyptian Arabic and says the drug and the test in English, in the same sentence. That's the hardest thing for speech-to-text systems. The first published test on exactly this kind of speech (Clinera, a preprint not yet peer-reviewed, March 2026) tried six systems on 683 medical sentences in Egyptian Arabic and English:
- The best system got 27.8% of the words wrong.
- Even that system produced 11 serious or potentially fatal medical errors, affecting 1.6% of the sentences.
- General multilingual systems barely recognised the medical terms at all.
In the UK, GPs said recording errors happen most in consultations that aren't in English (31%). A 2026 study in Riyadh of an Arabic-English system rated the quality of its Arabic notes highly, but it didn't measure the word error rate, didn't compare dialects, and most of its authors are linked to the company that makes the system.
What this means for your clinic: what comes out of the recording is a draft the doctor checks line by line, numbers first (blood pressure, weight, gestational age, doses). Nothing goes into the patient's file before the doctor signs it.
How much time does it really save?
Less than the ads promise, and more than nothing. The largest study so far (JAMA, April 2026, more than 8,000 clinicians at five academic health systems) found that those who used recording saved 16 minutes of writing for every eight hours of scheduled visits, and their time on records after clinic hours didn't change. In a randomised trial at UCLA (NEJM AI, 2025), one system cut note-writing time by 9.5%, and another made no meaningful difference.
The clearest effect wasn't on time but on fatigue. In a University of Wisconsin trial, work exhaustion went down, and in a study across six health systems, the share of clinicians reporting burnout fell from 52% to 39% within a month. For an OB/GYN clinic seeing thirty patients in an evening, that may be the real gain: your eyes stay on the patient, and the day ends without a report still owed to anyone.
Where does the audio go after the visit?
This is what patients in New York asked before anything else, and it's what you should ask before choosing a system:
- Which country is the recording processed and stored in? (In Egypt, storing it abroad is a cross-border transfer with its own conditions and permit.)
- When is it deleted? Once the report is checked, after some days, or never? (UC Davis deletes it within ten days.)
- Who besides the doctor hears it? Is it used to train the system?
- Does it stop the moment you stop it, and resume without losing what came before?
- Does it write into your clinic's own sheet and fields, or into a general template you then have to copy from?
Here's how we're building voice in Jyna: what you say becomes your clinic's own sheet, with its fields for each service, as a draft the doctor checks and signs. Nothing enters the file before that. The recording is deleted when the doctor signs, unless the doctor chooses to keep it.
Questions
Is it allowed to record a consultation by voice in Egypt?
We don't know of any text that forbids it, but the data protection law treats health data, which a recording is full of, as sensitive data that needs explicit written consent, and the Code of Medical Ethics protects the patient's secrets. Ask for consent before recording, write it down, and ask your lawyer about the form.
Does it understand Egyptian Arabic mixed with English medical terms?
It does, with errors. The best system in the first published test on this kind of speech got about one word in four wrong. That's why what it produces is a draft the doctor checks before it goes into the file.
What if the patient says no to recording?
The doctor finishes the visit and writes as usual, with a pen or a keyboard, and nothing about her visit changes.
You talk, and your clinic's sheet fills in. You're the one who signs.
In the demo we'll show you how what's said in a visit becomes your own sheet with its fields, and how you check it before it goes into the file.
Sources
- Leiserowitz G. et al., Patient Attitudes Toward Ambient Voice Technology, JMIR Medical Informatics, 27 November 2025, e77901; and the UC Davis Health news release, 16 December 2025.
- Lawrence K. et al., patient consent to ambient AI scribes, JAMA Network Open, 2025;8(7):e2522400.
- Thompson S., “I Felt Violated”: What One Patient Complaint Taught Me About AI in the Exam Room, Physician Leaders (AAPL), 18 May 2026, physicianleaders.org.
- General Medical Council, Making and using visual and audio recordings of patients. NHS England, Guidance on the use of AI-enabled ambient scribing products, 2025, updated 2026.
- Behairy A., Ahmed H., Clinera ASR Benchmark (Arabic-English, ARZ-EN), JMIR Preprints 96222, 28 March 2026, not yet peer-reviewed.
- Blease C. et al., survey of UK GPs on ambient scribes, BMJ Health & Care Informatics, 2026;33:e101847.
- Khan U.T. et al., A Bilingual Arabic-English Ambient AI Scribe, JMIR Medical Informatics, 24 March 2026, e83335.
- Rotenstein L.S. et al., JAMA, 1 April 2026;335(16):1408–17; and STAT News's coverage the same day. Lukac P.J. et al., and Afshar M. et al., NEJM AI, 26 November 2025. Olson et al., JAMA Network Open, 2 October 2025.
- Egypt's Personal Data Protection Law No. 151 of 2020, articles 1 and 12. The Code of Medical Ethics, Minister of Health Decree No. 238 of 2003, article 30.