Telehealth
The face is already on camera for the whole appointment. The vital sign is essentially free.
Patient monitoring
Contactless vitals
Vital signs from a camera your patient already owns. Pulse and respiratory rate recovered from ordinary video, with no cuff, no clip, and no device to ship, charge, clean, replace or chase.
01 · From a face
Contactless, a webcam is enough
Sit still in reasonable light for thirty seconds. Four independent algorithms read the same video and each reports its own number and its own confidence, because four estimates that agree tell you something a single estimate never can.
The face detector is fetched once from a public code CDN. Your video stays on your device throughout. Needs a camera and an https page.
02 · From a fingertip
Contact, this one needs a phone
Cover the rear lens and the flash with a fingertip. This is the same physics as the clip on a hospital finger, and a far stronger signal, strong enough to time the gap between individual beats.
Rear camera and flash required, so Android Chrome or Edge works best. Nothing is downloaded, because all of it ships with the page.
Remote monitoring has a reach problem,
not an accuracy problem
Device-based programmes work beautifully for the patients who receive a device, charge it, keep it and use it. That is a fraction of the population a health system is accountable for. Everyone else gets monitored by whatever they happen to report at their next appointment, which amounts to no monitoring at all.
A camera gives you a vital sign at zero marginal cost, on hardware the patient already owns. It can therefore be taken far more often, and by far more people, than a clinical-grade device ever will be.
Reach is the advantage here. Everything on this page follows from that.
Recovered signal four beats
What is measured
Skin colour, beat by beat
Every heartbeat pushes blood into the skin and changes its colour by an amount far too small for anyone to see. A camera can see it. All of the signal processing runs on the device, so no video is uploaded and the hardest conversation in remote monitoring is settled before it starts.
What the platform does
A measurement layer you embed into the patient-facing surface you already run, delivered as a component rather than as a hardware programme.
| Contactless capture | Pulse and respiratory rate from a few seconds of ordinary video of a face. Nothing is attached and nobody has to be approached, which is the entire point in a room you would rather not enter. |
| Fingertip capture | A finger over the rear camera and flash gives a far stronger signal, using the same physics as the clip on a hospital finger. It is strong enough to time the interval between individual beats. |
| Entirely on-device | Signal processing runs in the browser on the phone or laptop the patient already has. No video is uploaded. |
| Single-tag integration | One script tag and one element in your existing patient app, portal or telehealth surface. It emits a structured reading to your application, and what happens to that reading is entirely your system to decide. |
| An explicit quality gate | Where the signal will not support a reading, the platform returns none and says why. A wrong number in a monitoring record does more damage than a missing one. |
Where it sits in the care pathway
This is a wellness and engagement layer that extends a monitoring programme to the people it currently cannot reach. The four boundaries below are deliberate product decisions, and we state them up front.
Read this part before the rest
- It complements cleared devices and does not replace them. Where a regulated reading is required, a regulated device is required. This platform is for wellness and informational use only. It is not a medical device, it is not clinically validated, and it must not be used for diagnosis or treatment.
- A rhythm flag is a screening prompt. An irregular result means one thing: obtain a proper trace and consult a clinician. We do not present it as more than that, and we do not put it in front of asymptomatic populations at scale, where the arithmetic makes almost every flag a false one.
- Equity is measured. The method rests on a colour change in skin, and published accuracy across skin tones is uneven. We treat that as an engineering target to be measured and reported on, and we publish what we find.
- It declines to guess. Movement, poor light or a bad capture produce no reading and a stated reason. A system that reports a number it cannot stand behind does more harm than one that stays silent.
Where it deploys
The value concentrates wherever contact is the expensive part: the device, the trip, the gowning up, the cleaning between patients, or the patient you were never going to reach with hardware at all.
Isolation
A reading without someone putting on protective equipment to take it, and without a device that then has to be decontaminated.
Post-discharge
The window where deterioration is most likely and contact is weakest, and where a daily check-in is worth far more than a monthly one.
Onboarding and triage
A number on day one, before hardware ships, or in the far more common case where hardware was never going to ship.
Robots on rounds
The camera is already there and already passing every bed. Sensing costs nothing further once the robot exists.
Population engagement
Reaching the cohort a device programme structurally cannot, which is usually the majority of it.
Which patients is your programme not reaching?
Telehealth, isolation, post-discharge, or a robot already doing rounds. If you have a population you cannot reach with devices, that is the conversation we want to have.