Sector · Medicine
Nothing on this page makes a diagnosis. What the glasses carry is another doctor
Fifteen cases, and almost all of them are the same idea: a specialist who is not in the room, seeing what the hands in the room see. One system navigates inside the body and has FDA clearance. One study measured the checklist. One hospital is writing a ban.
Start with what is not here. No case on this page shows glasses reading a lesion and naming a disease. The medical use that actually exists is older and less cinematic: the glasses carry a second pair of eyes into a room where the specialist could not travel, and that is worth more than it sounds.
It started in 2013, when Dr Rafael Grossmann streamed a feeding tube placement at Eastern Maine Medical Center through Google Glass, deliberately choosing a simple procedure and broadcasting nothing that identified the patient. Everything since is that idea at scale. Proximie's platform, with 4K glasses over Vuzix hardware, serves more than 500 providers and 16,000 users in over 50 countries, including cases in Peru and Gaza, and records every procedure. Ohana One connected more than 100 surgeons across 5 specialties in 835 calls by June 2021 and now works in 35 countries, with the mentor able to put a hand into the trainee's field of view. At Chi-Mei Medical Center in Taiwan, an emergency team consulted a cardiac surgeon at another hospital about an aneurysm rupture, and the glasses streamed up to 16 hours without a break. In China, Sir Run Run Shaw Hospital runs an AR telemedicine network out to island clinics in Putuo district, and Qinghai's Haixi State People's Hospital broadcast nine complex operations live to eight municipal hospitals with synchronised annotation. Coverage figures for the Chinese network come from the supplier.
The one that goes inside the body
Augmedics xvision is the exception on this page: not a camera pointed outward but a navigation system, FDA cleared, that shows the patient's 3D anatomy so a spine surgeon places screws while looking at the patient instead of at a monitor across the room. The first US case was at Johns Hopkins. Cumulative figures cited in November 2025: more than 12,000 patients and more than 65,000 pedicle screws across 25 states.
What was measured, and by whom
Stanford and Columbia ran a surgical safety checklist on Google Glass against poster or memory across 80 preoperative time-outs at a large US teaching hospital. Completion reached 100% with the glasses against 76% conventionally, and the checklist ran 18% shorter. The authors also report the cost: the surgeon was isolated from the rest of the team while wearing it.
Two care numbers on this page are vendor case studies, and are labelled as such: a 50% increase in senior podiatry reach at Provide Community in Essex, and, at Uniting in Australia, 9 hospitalisations avoided and roughly 75% less referral time in four to six weeks with only four devices. The Hangzhou emergency triage trial that used Rokid Glass 2 during the Asian Games analysed 80 patients, of whom only 10 wore the glasses arm; that is a formative study, not evidence of practice. The USC team that used consumer Ray-Ban Meta glasses across 18 months of limb preservation surgery reported no device failures and one honest number: battery life fell from about 6 hours to 3 or 4 under heavy use. And the open source hospital ward scenario on GitHub is a planned use case in a project whose own documentation is 45% complete; there is no clinical deployment behind it.
The hospital writing the rule
Erie County Medical Center in Buffalo found it had no policy at all on consumer smart glasses in clinical areas, and started drafting one. The concern is not the surgeon wearing them on purpose. It is that a patient or a member of staff cannot tell whether the visitor across the room is recording, because the camera gives no obvious signal, and stickers sold online cover the indicator LED on the models that have one.