A teleprompter for patient education videos — clinical wording that does not drift

A patient education video is scripted for a reason. The aftercare for an extraction, the loading dose, the contraindications on a new prescription — these are sentences where the difference between “72 hours” and “a few days” is a phone call to the practice, or worse. Most clinical wording has already been through somebody: a governance lead, an indemnity insurer, the colleague who checks every patient-facing leaflet. The video has to say what the leaflet says. And the tenth time in a week you explain the same procedure from memory, the wording drifts — not because you know it less, but because you know it too well to notice.

Reading visibly makes it worse, not better. A patient watching your eyes track down a page beside the lens does not hear a careful clinician; they hear a recital. Warmth is the entire reason to make the video instead of handing over the leaflet — the point is that it feels like their own dentist or their own physio talking to them, not a document read aloud. Lose the eye contact and you have produced an expensive leaflet. Physios feel this hardest: half their videos demonstrate movement, and you cannot demonstrate a lunge while reading a page taped to a tripod.

And a clinical setting carries a question most video makers never face: where does the audio go? A teleprompter that streams your voice to a server for recognition is a conversation with your information-governance lead before it is anything else. The script names conditions and drugs; the recording is made inside your practice, between appointments. “Does this app upload anything?” is a question you will be asked in writing, and “I am not sure” is not an answer that survives an audit.

The usual method is memory, because clinicians know their material cold. But knowing the material and delivering it to a lens are different skills, and the retakes come from precision rather than ignorance — the ibuprofen and the paracetamol swapped in the sequence, “twice daily” said as “daily”, the contraindication remembered one sentence too late and awkwardly bolted on.

The fallbacks are a printed sheet taped beside the phone, or a practice nurse holding cue cards, both of which put your eyes visibly off the lens for every fact that matters. Dictation-style apps that could help with the reading tend to do their processing in the cloud, which puts them straight back into the governance conversation.

All of it happens in the one quiet room the practice has, booked into the gaps between appointments — which is how a two-minute video routinely eats forty-five minutes of a clinical day, and why most practices record one video, once, and never update it.

The script scrolls as you speak, and the speech recognition driving it runs on the device — your voice is processed on the phone and never uploaded. Pause to hold up the inhaler and demonstrate the technique, and the script waits. Skip a sentence, and it catches up. The approved wording stays in front of your eyes the whole way through, so “72 hours” is said as 72 hours on the first take rather than the fourth.

Cover mode puts the script over the whole camera view, so you are reading at the lens rather than beside it — the patient sees eye contact, and you are spared watching yourself while you work. For clinicians who dislike being on camera, which is most of them, this is often the difference between recording the video and postponing it again.

The script library holds one script per procedure — extraction aftercare, warfarin counselling, week-one ACL rehab — in folders, with a word count and an estimated read time on each, so you know before you book the room whether the script fits the gap. When the guidance changes, you edit the script once, and the next recording says the new wording; the old video is superseded rather than half-remembered.

Captions come from the script itself, so the subtitle file carries the wording you approved instead of an auto-transcriber’s guess at a drug name, and you can export them as a separate .srt for the practice website or the patient portal’s player. Delivery analysis runs on the device too, and recordings stay in the app’s own storage until you choose to export one. One honest note: exports on the free tier carry a small Talk2Camera mark; a practice that wants unmarked video will want the paid plan.

A dentist has twenty minutes between the morning list and lunch. The extraction-aftercare script is already in the library — 340 words, a shade over two minutes on the estimate. She props the phone on the desk in the consulting room, puts the script over the camera, and reads it to the lens: the bleeding, the rinsing, the smoking, the painkiller sequence in the approved order. Halfway through she picks up a gauze pack and shows how to fold and place it; the script waits until she speaks again.

One take. She exports the video with an .srt for the practice website, where most patients will watch it muted in a waiting room or on a bus home, and the link goes out with the appointment follow-up. The next patient who rings about dry socket has already heard the same wording the leaflet uses — because it is the leaflet’s wording, read warmly, to a lens.