Does Your Wearable Actually Measure Respiratory Rate Accurately?

This piece covers what published validation research shows about respiratory rate estimation from wearable and contactless sensors (PPG, ECG, accelerometer, stretch sensors) compared to reference medical devices. It does not cover diagnosis of specific respiratory conditions or advice about interpreting your own readings.

Research on wearable respiratory rate monitors shows the technology can track breathing rate closely enough to agree with medical-grade reference devices in controlled comparisons, but the accuracy depends heavily on which sensor type, body location, and condition (rest versus movement, illness versus health) is being tested. Chest-worn and photoplethysmography-based devices have shown strong agreement with clinical references in specific validation studies, while broader reviews across many consumer and contactless devices find the evidence base uneven and method-dependent.

The gap between a heart rate number and a breathing rate number

There's a specific kind of doubt that shows up once people notice their wearable displays a respiratory rate at all. Heart rate feels intuitive, you can feel your own pulse. Breathing rate, measured from a wrist or a ring, feels like it's being inferred rather than sensed, and that makes people wonder whether the number is a real measurement or a rough guess dressed up as data.

That skepticism isn't unreasonable. Respiratory rate on most consumer wearables isn't measured directly the way a nasal airflow sensor would measure it. Instead it's typically extracted from secondary signals, tiny fluctuations in your pulse waveform or heart rhythm that happen to be modulated by breathing. Whether that indirect route holds up under scrutiny is exactly what the validation research was built to test.

What the comparisons against medical-grade devices actually found

The most direct answer comes from studies that put a wearable side by side with a reference standard, like a capnograph, ventilator monitor, or nasal cannula, and checked how closely the numbers tracked. One clinical validation of a photoplethysmography-based wearable ran three separate comparisons: against a capnograph in healthy volunteers, against a ventilator monitor in ventilated patients, and against a capnograph again in COVID-19 patients with active pulmonary disease. Correlations were high across all three settings, and the bias between the wearable and reference device stayed under one breath per minute in each case. The spread of individual differences, though, was wider in the sicker patient group than in the healthy volunteers.

A separate study of two wearable stretch sensors placed at different points on the torso, chest, abdomen, epigastrium, found that sensor placement mattered: the strongest signal amplitude showed up at the epigastrium and umbilicus depending on whether the person was sitting or lying down. That's a detail easy to miss if you assume a respiratory sensor works the same no matter where it sits on the body.

Why the broader research picture is less tidy than a single study suggests

Zoom out from individual devices to the systematic reviews, and the picture gets more cautious. One meta-analysis pooling studies of wearable and contactless devices measuring heart rate, respiratory rate, and oxygen saturation in clinical settings found that wearable device studies were more often commercially available products tested in acute clinical settings by clinical staff, with more real-time data analysis, while contactless devices tended to be more experimental and analyzed after the fact. The review also flagged a higher risk of patient selection and rater bias in contactless device studies compared to wearable ones.

A separate review focused specifically on consumer-grade contactless monitors, the kind using a regular camera rather than a dedicated sensor, found that only a small slice of the included studies even measured respiratory rate at all, most were testing heart rate. Where respiratory rate was assessed, the review noted limitations tied to motion and other interference. Neither review is describing wrist-worn PPG devices specifically, but together they show the strength of evidence differs quite a bit depending on whether you're looking at a wearable or a camera-based contactless system.

The underlying signal-processing research backs this up from a different angle. A review of algorithms that estimate breathing rate from ECG and PPG signals describes dozens of proposed methods, and a separate systematic algorithm comparison found that performance varies depending on which extraction technique is used and whether the signal comes from ECG or PPG. The sensor is only half the story; how the raw signal gets converted into a breaths-per-minute number is the other half, and that variability at the algorithm level is part of why device-level accuracy isn't uniform.

Where accelerometers and unconventional sensors fit in

Not every respiratory rate estimate comes from a pulse signal. One study combined an ECG-derived respiratory signal with data from a chest-worn accelerometer, on the logic that breathing subtly shifts the whole ribcage, and found a mean absolute error of about one breath per minute during steady metronome-paced breathing, rising during activities of daily living. Another looked at whether the accelerometer and gyroscope inside a virtual reality headset, originally there for head tracking, could pick up respiratory movement well enough to estimate rate. The gyroscope won out over the accelerometer for that purpose.

There's also work using wrist PPG signals for something beyond a simple rate number: one study extracted respiratory rate from post-operative patients' wrist PPG specifically to help classify pain levels, since breathing patterns shift with pain. That's a reminder that respiratory rate on a wearable isn't just a standalone vital sign, it's increasingly treated as an input feeding other estimates. For anyone comparing how these numbers get used across contexts, it connects to the broader question of what a single respiratory rate reading is actually good for once it leaves the lab.

What a review of contact-based methods says about the ceiling here

A wide-ranging review of contact-based respiratory rate methods covers the full menu of approaches, airflow, sound, temperature, humidity, chest wall movement, and modulation of cardiac signals, and lays out the working principles and metrological characteristics behind each. It doesn't crown one method as universally best. It frames the choice as dependent on the application, whether that's clinical monitoring, occupational settings, or exercise, and that framing pushes back on the idea that there's one correct way to measure breathing rate on a wearable. Chest strain sensors, wrist PPG, ECG-derived estimates, and accelerometer-based methods are all legitimate approaches with different strengths, and the review that assessed algorithms systematically found that no single combination of extraction technique and signal source dominated across the board.

Common questions