TL;DR. A gait screen lasts eight seconds and you are usually cueing the patient while it happens, so the moment to press record has already gone by the time you have a hand free. Mount the phone once, start the buffer, run the screen. A tap on your watch keeps the pass that mattered, and you step through mid-stance one frame at a time while the patient sits down. Patient footage never leaves the device, because there is no account and no server to leave it to.
ReplayR is a camera, not a medical device. It records and plays back video. It does not measure joint angles, diagnose, or produce clinical output, and nothing on this page is medical advice.
Sports video and clinic video look like the same job and are not. On a sideline you have one hand free, a wide space, and footage of people who came to play a match. In a treatment room you have neither hand free, four metres of floor if you are lucky, and footage of a patient, which is a different kind of file with different obligations attached to it.
Three practical problems follow, and they are what the rest of this page is built around.
During a step-down or a single-leg squat you are guarding, cueing, or with a hand on the pelvis. There is no moment where you can also be a camera operator. Anything that needs you to reach for the phone before the movement will be missed, every time, until you stop bothering.
A 10 metre walk takes eight to twelve seconds. Five squats take fifteen. By the time you have decided the pass was worth keeping, the pass has finished. A buffer that was already running is the only version of this that works, because the decision comes after the event rather than before it.
A clip of a patient's gait is identifiable health-related material. Where it lives, who processes it and how long it is kept are questions your practice has to answer. An app that uploads to a vendor's cloud by default turns a five-second recording into a data processing arrangement.
The buffer answers the first two. The absence of a server answers the third. ReplayR keeps a rolling window of the last 45 seconds to 12 minutes of video, overwrites everything older, and writes a clip to disk only when you tell it to. There is no account, no login and nowhere for footage to go except the phone in the room.
Camera position is the one thing worth being fussy about. Everything else forgives a rough setup. A badly placed lens gives you footage where the finding is hidden behind the patient's own body, or behind you.
Mark the tripod position on the floor with tape and write the height on it. Six weeks later, the reassessment is only meaningful if it was shot from the same place, and the difference between "the knee looks better" and "the knee is better" is often just whether the camera moved. Work out exact distances for your own room with the camera angle and distance planner.
| What you are filming | Typical length | Buffer to set | Tier |
|---|---|---|---|
| 10 metre walk, one direction | 8–12 s | 45 s | Free |
| Walk out, turn, walk back | 20–30 s | 45 s | Free |
| Five bodyweight squats | 15–25 s | 45 s | Free |
| Step-down or single-leg squat, both sides | 40–70 s | 90 s | Free |
| Hop test battery, one limb | 1–2 min | 3 min | Pro |
| Treadmill running gait, settled pace | 1–3 min | 3 min | Pro |
| Full multi-item movement screen, unbroken | 6–10 min | 12 min | Pro |
| A rehab exercise set, mid-session | 30–60 s | 90 s | Free |
Most clinic work sits comfortably in the free 45 and 90 second windows, because a screen item is short by design. The reason to go longer is a whole battery you want intact in one file, or a treadmill assessment where the interesting stride happens once the patient has stopped thinking about being filmed. Check what a given buffer costs on your device with the storage and buffer time calculator.
This is the part that decides whether video survives past the second week in a real caseload. Twenty steps to the phone and twenty back, per item, per patient, and it quietly stops happening. Two triggers remove the walk.
A Tile or a watch-face Complication writes the last window to disk in a few hundred milliseconds. You do it as the patient finishes the pass, without moving, without breaking eye contact, and without announcing to the room that something is being recorded. The watch also starts and ends the session and changes clip length, so the phone can stay on its tripod from the first patient to the last. The link is Bluetooth, so a clinic with no usable network behaves the same as one with fibre.
Hold a flat open hand towards the lens for a moment and the clip saves itself. Useful when you are not wearing a watch, or when your hands have just come off a patient and you would rather not touch a phone screen before you have washed them. It fires from across the room, so the phone stays where you put it.
The buffer keeps rotating through a save, so you can trigger twice in ten seconds and get two clips. Between patients, the swipe review screen is where the session gets cleaned up: keep the two clips worth keeping, throw the rest, and what you threw sits in a bin with a 24 hour recovery window in case you were too quick.
Stealth mode is worth turning on for clinic use specifically. It blacks the screen while the buffer runs, which stops a bright live preview of the patient facing into a shared gym, and stops the patient watching themselves walk, which changes how they walk. Black screen, buffer running, watch still working.
Slow-motion playback down to 0.25x and the trim editor are on the free tier and handle the between-items look. That alone changes the assessment, because the events you care about last well under a tenth of a second. Initial contact, the instant of peak knee valgus, the frame where the pelvis drops: at normal speed these are one blurred frame and a memory of what you thought you saw.
Pro adds the analyzer: step forward and back one frame at a time, hold to run, scrub frame-precise, and drop to 0.1x. Full detail in the frame-by-frame guide.
Worth being clear about the limits of this. What you have is a single camera in one plane at consumer frame rates. That is honest observational video, and it is a large improvement on unaided eyes, but it is not kinematic measurement. Reported angles from 2D video depend heavily on camera alignment and on where the observer decides a landmark is, which is exactly why marker-based 3D systems still exist. Use the video the way you use your own observation, as one input into clinical reasoning rather than a number to build a plan on.
Half the value of clinic video is not diagnostic at all, it is explanatory. A patient who has been told for three visits that their knee falls inwards, and who has never seen it, is being asked to take your word for a sensation they cannot feel. Thirty seconds of playback at 0.25x usually ends that conversation, and it tends to be the visit where adherence to the home programme changes.
On Pro, once the analyzer is on the frame you want, tap Annotate frame and draw on it: freehand, straight lines and short text labels, four colours and three sizes, undo and clear-all. A vertical dropped from hip to floor through the knee at mid-stance makes valgus impossible to argue with. A horizontal across the iliac crests shows a drop that the patient cannot feel. The export is one flattened image at the clip's recording quality, so the marks are part of the picture. The annotation walkthrough covers the tools in full.
The habit that pays off: export the same frame at assessment and at reassessment, same camera position, same line drawn on it. Two images side by side settle the question of whether six weeks of work changed anything, which is otherwise a matter of opinion between you and the patient. It also gives you something concrete for a discharge summary or a referral letter that a paragraph of prose does not.
Marks are drawn on a still, not tracked across the moving clip, and no angle is calculated for you. That is a deliberate limit rather than a missing feature: a drawn line is a teaching aid, and treating it as a measurement would be a mistake.
This is the section most video analysis apps skip, and it is the one that decides whether a clinic can use the thing at all. Video of an identifiable patient, recorded in the course of their care, is health-related personal data. In the United States that puts it inside HIPAA when it is held by a covered entity. In the UK and EU it is special category data under UK GDPR and the GDPR, which means a lawful basis and an Article 9 condition, not just a signature.
No account, no login, no server, no analytics on the footage. A clip exists on the phone that recorded it and nowhere else. There is no vendor to sign a business associate agreement with over patient video, because there is no vendor holding patient video. That removes a category of paperwork rather than answering it.
The flip side of on-device storage is that the phone's own security is the whole security model. Screen lock, device encryption, a clinic-owned handset rather than a personal one, and a rule about what happens if it is lost. If the footage forms part of the clinical record, it needs to reach your records system and be retained there under your normal policy, not left in a camera roll.
Say what you are recording, why, who will see it, how long it is kept, and that they can decline without it affecting their care. Record that you asked. Consent for assessment is not consent for teaching, and neither is consent for a social media post. Ask again, separately, each time the purpose changes.
The swipe review at the end of a session is the natural moment. Keep what goes in the record, bin the rest, and let the 24 hour recovery window cover mistakes. A retention rule that depends on someone remembering to tidy a phone in six months is not a retention rule.
Shared gyms need one extra thought: frame tight on your own patient rather than wide across the room, which keeps other people out of shot as a side effect and gives you a better picture anyway. The guide to recording ethics in sports facilities goes into the consent and bystander questions in more depth. None of this is legal advice, and your professional body's guidance and your practice's information governance policy take precedence over anything here.
Clinical video analysis platforms such as OnForm, and the telehealth-oriented tools around them, solve a different problem: shared libraries, per-client folders, voice-over feedback, remote review, side-by-side comparison, and a per-seat subscription to pay for the hosting that all of that requires. If your practice runs remote programming and needs a client to receive annotated video between visits, that is what those products are for, and this is not a replacement.
The gap they leave is capture. They assume you already have the clip, which puts you back at the problem of pressing record with no hands during an eight second walk, then trimming a four minute file down to the useful eight seconds afterwards. ReplayR is the capture half: the buffer was already running, the watch tap keeps only what mattered, and the file that lands is already the right length. Costs $5.49 once, per device, with no seat count and no renewal. Full breakdown in ReplayR compared with OnForm.
Saved clips are ordinary MP4 files. If you do run an analysis platform, handing a clip over is a normal share, and the buffer just means you are handing over eight useful seconds instead of four minutes of a mostly empty room.
Filming everything is the fastest route to reviewing nothing. A full caseload generates dozens of near-identical clips, and a folder of dozens is a folder nobody opens. Sample instead.
Two or three clips per patient, reviewed in the room while it is still relevant and compared at reassessment, beat forty clips reviewed never. Because saving costs a wrist tap, the price of following that rule is close to zero, which is the only reason anyone keeps it past the second week.
No. There is no account, no login and no server. Clips are written to the device that recorded them and stay there until you delete them or share them yourself. That removes the whole category of questions about which vendor holds the footage, in which country, and for how long. It also means the device's own security is the only thing protecting the file, so a screen lock, full-device encryption and your clinic's mobile device policy are doing the real work.
Start the buffer once at the beginning of the session and leave the phone on its tripod. It keeps a rolling window of the last 45 seconds to 12 minutes and overwrites the rest. When a pass is worth keeping, a tap on a paired Wear OS watch writes it to disk, or on Pro you hold an open palm towards the lens. Neither requires walking to the phone or interrupting a hands-on assessment.
45 seconds covers a single item: five squats, a set of step-downs, one 10 metre walk. 90 seconds holds a walk-out and walk-back with the turn included. A full seven-item screen without stopping runs 6 to 10 minutes, which needs a Pro buffer. Both 45 and 90 seconds are on the free tier, and most clinicians never need more because a screen item is short by design.
No. It is a camera with a rolling buffer and a slow-motion player. It measures nothing, calculates no joint angles, produces no report and makes no clinical claim. Kinematic output, marker tracking and validated angle measurement are the domain of 3D motion capture systems and of software cleared for that purpose. Treat what you see here the way you treat what you see with your own eyes, as observational data feeding your clinical reasoning.
On Pro, yes, on a still frame rather than on the moving clip. Step the analyzer to the frame you want, tap Annotate frame, then draw a vertical through the knee at mid-stance or a horizontal across the pelvis to show a drop. Saving flattens the marks into one image at the clip's recording quality. A single annotated still is usually a better teaching object than a video, and it is small enough to hand over without a file transfer problem.
Yes, and clinic basements and hospital gyms are exactly where cloud-based analysis apps fail. Nothing here touches a network. The watch link is Bluetooth, capture and playback are local, and there is no sign-in screen between you and the buffer. A dead spot changes nothing about the workflow.
No. The buffer occupies a fixed block of storage and overwrites its own oldest frames, so eight hours costs what ten minutes costs. Only clips you deliberately keep accumulate. The swipe review at the end of a session clears the rest into a bin with a 24 hour recovery window, which is also the point in the day where a clinic policy of deleting patient footage after review is easiest to actually follow.
Governance and professional guidance referenced above. Verify against your own jurisdiction and professional body before setting a clinic policy.
Last verified . Not legal or medical advice.
All apps on Google Play
