Does Wrist vs Chest Placement Affect Respiratory Rate Accuracy?

This article covers research comparing body placement (wrist, finger, forehead, neck, chest/torso, head) for deriving respiratory rate from wearable sensors, primarily PPG and inertial motion sensors. It does not cover clinical-grade capnography or nasal airflow devices, and it does not address exercise-specific accuracy.

Placement changes respiratory rate accuracy because different body sites pick up breathing through different physical pathways, chest and abdomen sensors detect the mechanical movement of breathing directly, while wrist and finger sensors have to infer breathing indirectly from how it modulates the pulse waveform. Research comparing sites like the finger, forehead, wrist, earlobe, and arm has found that the strength of these indirect respiratory signals varies by location, which is a measured, site-dependent difference rather than a settled ranking of one wearable form factor over another.

Why the same breath looks different depending on where the sensor sits

It is a fair question to ask why a wrist-worn device and a chest strap might give different breathing numbers for the exact same few minutes of resting on a couch. Nothing about the person changed. What changed is what the sensor is actually touching.

A chest or abdomen band feels breathing the direct way, as the physical rise and fall of the torso wall. A wrist or finger sensor typically relies on photoplethysmography (PPG), a light-based pulse signal that breathing modulates only indirectly, through small changes in blood volume, timing, or pressure as the chest moves. That indirect path is where placement starts to matter.

What happens away from the wrist and chest

Some of the more interesting placement research does not involve the wrist or chest at all. A head-worn virtual reality mask with a forehead PPG sensor has been used to estimate breathing rate with a machine-learning approach, treating the forehead as its own viable location rather than a stand-in for the wrist. Motion sensors built into VR headsets have also been tested for extracting respiratory rate from head movement itself, using gyroscope and accelerometer data compared against a controlled-breathing reference across sitting, standing, and lying-down postures.

The neck has been explored too, combining PPG with accelerometer data to estimate both pulse rate and respiratory rate, despite the neck being described in that research as an area highly prone to motion artifact. And IMU-based systems placed directly on the thorax, abdomen, and lower back, similar in spirit to general accuracy questions about wearable respiratory rate, have been shown to detect statistically significant differences in respiratory rate between static postures and dynamic activities like walking, running, and cycling.

What ties these together is not a verdict on which single spot is correct. Each site is being asked to solve the same problem, extracting a slow breathing rhythm from a signal built for something else, whether that is pulse, motion, or skin surface stretch.

Reading placement studies without over-reading them

It is tempting to turn a study that found finger PPG outperforming forehead PPG into a general rule that fingers beat everything, or to turn a wrist-worn validation into proof the wrist is fine for all purposes. The research does not support either leap. A large review of contact-based respiratory rate methods lays out airflow, sound, temperature, humidity, chest-wall movement, and cardiac-activity-modulation approaches as genuinely different measurement families, not variations on one theme, which is part of why placement interacts with method in ways that resist a single ranking.

A broader review of breathing rate estimation from ECG and PPG signals frames this as an open methodological area, noting that despite considerable work, using these algorithms reliably in wearable sensors and clinical practice still has open questions. Systematic algorithm comparisons using ECG and PPG signals against a reference respiratory measure have found performance depends heavily on which signal-processing stage and technique combination is used, independent of body site.

None of this settles into wrist-is-worse or chest-is-better. It settles into something plainer: the signal pathway from a given body site interacts with the extraction method, and both matter more than the wearable's form factor alone.

Where this leaves the wrist-vs-chest question specifically

Direct head-to-head wrist-versus-chest comparisons are thinner in this evidence set than comparisons involving finger, forehead, earlobe, arm, and neck placements. What is measurable from the studies available is that wrist PPG has been used successfully as one location among several in respiratory rate validation work, and that indirect PPG-based approaches generally carry a different accuracy profile than direct chest or abdominal movement sensing, which several IMU-based systems targeted explicitly.

A stretchable, non-contact morphic sensor embedded in a shirt, compared against overnight polysomnography, reached close agreement with the gold standard for computed respiratory parameters, suggesting torso-based direct movement sensing can perform well in a sleep-monitoring context specifically. Whether that gap holds up, narrows, or reverses in daily wrist-worn use during waking hours isn't something this set of studies directly answers, which connects to the wider question of whether wearable respiratory rate tracking can flag meaningful health changes early.

Common questions