Can an App Measure Your Blood Pressure?

Blood pressure is a force. Every method of measuring it, from a mercury column to your automatic cuff, works by opposing that force with a known pressure and detecting when blood flow stops or resumes. The cuff is not a packaging choice — it is the measuring instrument.

A phone pressed against a fingertip has no way to apply or sense a known counter-pressure. What the camera can see is a photoplethysmogram: the light through your fingertip brightening and dimming with each pulse. That signal genuinely contains a heart rate, and it carries some information about your pulse waveform. What it does not contain is a scale — nothing tells the phone whether that waveform belongs to someone at 110/70 or 160/95.

Cuffless wearables get around the missing scale by calibrating against a real cuff, per person, and recalibrating as the relationship drifts. An app that has never seen a cuff reading of yours has skipped the only step that makes the estimate possible. However good the algorithm, it is guessing from population averages.

The FDA has issued a safety communication telling people not to use blood pressure devices that have not been authorized — devices which have not been reviewed for safety and effectiveness. The risks it names are not abstract: failing to detect a genuine change in your blood pressure, and delaying treatment because a reassuring number arrived from an app.

Research has been blunter. Studies of finger-on-the-screen apps have found readings that both miss most cases of high blood pressure and reassure people who have it. One widely reported app was independently tested against a validated monitor and got the great majority of hypertensive readings wrong in the reassuring direction — a specific and dangerous failure mode, because the person most harmed is the one who most needed to know.

The distinction to hold on to: there is serious research into estimating blood pressure from optical signals, including phone-based approaches under clinical trial. Those are investigational, regulated, and evaluated against reference measurements. That is a different thing from a consumer app with no clearance shipping a number to millions of phones today.

• It never asks for a cuff reading. No calibration means no per-person mapping. This alone is decisive.
• It measures through the camera, screen or fingerprint sensor. No consumer phone has hardware that can measure blood pressure.
• The numbers are too well-behaved. Real blood pressure varies noticeably between readings. Values that hover near 120/80, or move only a little, suggest a generator rather than a measurement.
• The store listing hedges. "For entertainment purposes," "not a medical device," or "estimates only" in the small print, under a screenshot showing a confident reading, tells you the developer knows.
• It claims blood glucose too. Non-invasive glucose from a phone camera is not a solved problem, and a listing offering it is not being careful about anything.

A useful test: ask what the app would do differently if you handed the phone to someone else. If nothing in its process is specific to you, its number cannot be either.

Everything valuable about a blood pressure app happens after a real monitor produces a reading — and that is not a consolation prize, because the interpretation is where most home monitoring falls apart.

• Averaging. Single readings are noisy and misleading. Guidelines are written on averages — the 7-day protocol exists precisely because one number means little. Arithmetic across a week is exactly what software is for.
• Capture without transcription. Typing numbers from a display invites transposition errors, and errors in a log are worse than gaps. A camera that reads the monitor's own screen removes that step.
• Context. Time of day, which arm, before or after medication, how you slept — the things that explain an outlier and stop you panicking about it.
• Classification you can trust. Color-coding each reading against the current guideline categories, rather than remembering where 134/86 falls.
• A report your doctor can read. A PDF of the whole log with averages beats a phone screen scrolled at an appointment.
• Consolidation. Readings from a Bluetooth monitor, a wearable and a manual entry all landing in one history rather than three apps.

There is a legitimate use of a phone camera in blood pressure, and it is worth being precise about, because it sounds superficially like the thing this guide just warned about.

Pointing the camera at your monitor's display and having software read the digits is optical character recognition. The measurement was performed by the cuff; the camera is replacing your typing, not the device. If the scan misreads a digit, you see it immediately against the screen in front of you.

Pointing the camera at your finger and having software produce a blood pressure is a claim to have measured you. There is no cuff anywhere in that chain.

Same sensor, entirely different claim. This is the line worth checking on any app that mentions a camera: is it reading a device, or is it claiming to read you?