Respiratory Rate From ECG and PPG: Signal Quality Rules for Wearable Breathing Metrics

Respiratory rate from ECG and PPG is an inferred breathing metric. ECG captures respiration through respiratory sinus arrhythmia and respiration-linked changes in QRS morphology. PPG captures respiration through amplitude, baseline, and pulse-rate modulation. The metric is reliable only when algorithms reject poor ECG and PPG segments before estimating breaths per minute.¹²

This distinction matters. Respiratory rate is clinically meaningful, but ECG and PPG do not measure airflow. They measure cardiovascular signals that respiration modulates. A wearable breathing metric therefore depends on physiology, sensor contact, motion context, artifact handling, and validation against a reference respiratory signal.³⁴

Sensor Bio treats these signals as infrastructure. Raw PPG, ECG-derived intervals, and quality metadata should remain accessible for research pipelines. See [Sensor Bio's validation methodology](/science/) and [photoplethysmography fundamentals](/the-signal/photoplethysmography-ppg) for the broader signal architecture.

Why respiratory rate matters

Respiratory rate is one of the core vital signs. It is also frequently undermeasured or manually counted with limited precision. Cretikos et al. described respiratory rate as a neglected vital sign because changes may precede visible clinical deterioration.³³ Fieselmann et al. found that respiratory rate predicted cardiopulmonary arrest in internal medicine inpatients.³⁴

Early warning systems include respiratory rate because it carries information that heart rate and blood pressure do not fully capture. The Modified Early Warning Score literature, centile-based warning models, and hospital deterioration studies all support respiratory rate as a high-value physiological variable.³⁵³⁶³⁷ Automated respiratory rate may also differ from manual ward measurement, which creates a measurement-method problem before interpretation begins.³⁸

Wearables make continuous estimation possible. They also introduce a new risk. A continuous respiratory rate stream can look precise while being artifact-dominated. The number can be mathematically stable and physiologically wrong. That is why signal quality rules must precede any clinical or research interpretation.

How ECG carries respiration

ECG-derived respiration uses two mechanisms.

First, respiration changes the timing between heartbeats. Respiratory sinus arrhythmia describes the cyclic acceleration and deceleration of heart rate across the breathing cycle. It reflects respiratory coupling with autonomic regulation, especially vagal modulation.⁴⁰⁴¹ During many resting conditions, heart rate tends to increase during inspiration and decrease during expiration. That rhythm creates a respiratory component in RR intervals.

Second, respiration changes the ECG waveform. Chest movement and heart-axis shifts can modulate QRS amplitude and morphology. Electrode position, thoracic impedance, and posture can alter the strength of this signal. ECG-derived respiration therefore often combines interval-based and morphology-based features.²⁴⁷

The physiology is useful, but it is not universal. Respiratory sinus arrhythmia weakens with age, medication effects, disease state, stress, and irregular rhythm. It also depends on breathing depth and rate. HRV standards and methodological reviews warn that autonomic interpretation requires careful attention to recording conditions, artifact correction, and respiration itself.⁴³⁴⁴⁴⁵

This creates a practical rule: ECG-derived respiratory rate is strongest when R peaks are clean, rhythm is sufficiently regular for interval analysis, and the respiratory component is visible in the expected frequency band. If ectopy, arrhythmia, motion artifact, or poor electrode contact corrupts the ECG, the derived respiratory estimate should be down-weighted or rejected.²⁹⁴⁸

How PPG carries respiration

PPG measures optical changes linked to blood volume in tissue. It is widely used for pulse timing, pulse amplitude, and waveform morphology. The signal contains cardiac pulses, slower vascular trends, and artifact from contact, motion, ambient light, and tissue properties.⁸⁹¹¹

Respiration appears in PPG through several respiratory-induced variations.

Respiratory-induced amplitude variation reflects changes in pulse amplitude across the breathing cycle. Respiratory-induced intensity variation reflects baseline or low-frequency shifts in the optical signal. Respiratory-induced frequency variation reflects respiratory modulation of pulse timing, often overlapping with pulse rate variability.¹⁰¹⁵

These mechanisms are physiologically plausible. Intrathoracic pressure, venous return, stroke volume, vascular tone, and autonomic coupling can all influence the peripheral pulse waveform. Meredith et al. reviewed the relevant physiology behind deriving respiratory rate from PPG.¹⁰ Karlen et al. showed that combining multiple PPG-derived respiratory signals can improve respiratory rate estimation compared with relying on a single modulation channel.³

The same mechanisms create vulnerability. Motion can change amplitude. Contact pressure can change baseline. Vasoconstriction can reduce waveform quality. Poor perfusion can weaken the pulse. A PPG algorithm can mistake non-respiratory amplitude fluctuation for breathing unless it gates by signal quality and cross-checks multiple features.⁶²⁶

This is where [PPG signal quality](/the-signal/ppg-signal-quality) becomes central. Respiratory rate from PPG is not just a spectral peak. It is a decision about whether the observed modulation is stable, physiologically plausible, and consistent across signal channels.

Why ECG and PPG fusion is the stronger architecture

ECG and PPG fail in different ways. ECG may preserve R-peak timing when peripheral perfusion is poor. PPG may preserve respiratory amplitude modulation when ECG morphology is noisy. Neither is sufficient by default.

The literature supports fusion because respiration leaves redundant but imperfect traces across cardiovascular signals. Nemati et al. described data fusion for improved respiration rate estimation.⁴ Birrenkott et al. developed a fusion model for respiratory rate from PPG and ECG.⁵ Khreis et al. used a Kalman smoother across ECG and PPG respiratory estimates.²² Adami et al. extended the framework across ECG, PPG, and blood pressure signals.²³

Fusion should not mean averaging everything. A simple average can make a bad estimate look credible. The better architecture estimates respiratory rate from multiple derived respiratory signals, scores each signal, rejects low-quality channels, and fuses only the channels that pass quality rules.²¹

This is also why raw and intermediate signals matter. A platform that exposes only a final respiratory rate hides the quality evidence needed to audit the metric. Research pipelines need timestamped ECG, PPG, derived respiratory traces, signal quality indices, and rejection flags. For modality differences, see [how PPG compares with ECG and pulse oximetry](/the-signal/ppg-vs-ecg-vs-pulse-oximetry).

Signal quality rules for wearable respiratory rate

A wearable respiratory rate pipeline should apply quality rules before, during, and after estimation. The rules below are methodological requirements, not product claims.

Rule 1: require clean cardiac fiducials

ECG-derived respiration depends on accurate R-peak detection. PPG-derived frequency variation depends on accurate pulse detection. If the pipeline cannot locate beats reliably, it cannot estimate respiration reliably from beat timing.²⁷²⁸

For ECG, reject segments with poor electrode contact, saturated amplitude, excessive baseline wander, or uncertain QRS detection. For PPG, reject segments with missing pulses, clipped peaks, unstable baselines, or morphology inconsistent with a physiological pulse. Beat-level uncertainty should propagate into respiratory-rate uncertainty.

Rule 2: separate respiratory modulation from motion artifact

Motion produces large low-frequency components in PPG. Those components may sit in the same frequency range as breathing. Wrist, finger, and ring positions each have contact-mechanics constraints. The algorithm must distinguish respiratory modulation from mechanical modulation.⁹²⁶

Accelerometer context can help. Stillness is not a guarantee of quality, and motion is not always fatal. The practical rule is segment-level evidence. If PPG amplitude modulation tracks movement more than pulse physiology, reject it for respiratory estimation.

Rule 3: score ECG and PPG independently

Signal quality is modality-specific. A clean ECG does not prove clean PPG. A clean PPG does not prove clean ECG. Orphanidou et al. derived signal quality indices for ECG and PPG in wireless monitoring.⁶ Clifford et al. showed how ECG signal quality and data fusion can determine clinical acceptability.⁷

Respiratory pipelines should keep separate quality scores for ECG R peaks, ECG morphology, PPG pulse detection, PPG amplitude stability, PPG baseline stability, and derived respiratory traces. A single global score hides failure modes.

Rule 4: require agreement across derived respiratory signals

PPG can generate multiple respiratory traces. ECG can also generate interval-based and morphology-based respiratory traces. Agreement across independent traces increases confidence. Disagreement should trigger down-weighting, wider uncertainty, or rejection.³¹⁵

Agreement does not require exact equality. Breathing is nonstationary. The rule is physiological coherence over a defined window. If one trace estimates 12 breaths per minute and another estimates 28 without a clear reason, the pipeline should not publish a single confident value.

Rule 5: constrain estimates to plausible physiology

Respiratory rate estimation is vulnerable to harmonics. Algorithms may lock onto half-rate or double-rate peaks. Time-frequency methods, correntropy spectral methods, and quality-index fusion all address this issue in different ways.¹²¹³²⁴

Plausibility checks should include age group, context, recent trend, breathing-rate continuity, and signal confidence. Sudden jumps can be real, but they need supporting signal evidence. A quality-aware pipeline should flag abrupt changes that appear without matching respiratory modulation.

Rule 6: validate against reference respiration

ECG and PPG are indirect respiratory sensors. Validation should compare estimates against a reference respiratory signal, such as capnography, respiratory inductance plethysmography, or another accepted respiratory measurement method. Algorithm papers commonly use reference datasets and waveform databases for this reason.^14303132

Validation should report error distributions, not only average error. Bland-Altman limits, coverage across activity states, missingness, rejection rates, and subgroup performance matter. A low mean error can hide poor performance during motion or low perfusion.

Evidence summary

| Evidence area | Study design or source type | Signal quality implication | Citation | |—|—|—|—| | ECG and PPG respiratory rate review | Methodological review | Breathing rate can be estimated from both modalities, but indirect mechanisms require validation | Charlton et al., 2018 ² | | PPG multiparameter estimation | Algorithm study | Multiple respiratory-induced variations outperform reliance on one PPG trace | Karlen et al., 2013 ³ | | ECG and PPG fusion | Algorithm and fusion studies | Fusion improves reliability only when noisy channels are scored and weighted | Nemati et al., 2010; Birrenkott et al., 2018 ⁴⁵ | | Signal quality indices | ECG and PPG quality-method papers | Quality scoring should occur before respiratory-rate interpretation | Orphanidou et al., 2014; Clifford et al., 2012 ⁶⁷ | | Clinical importance of respiratory rate | Vital-sign and early-warning studies | Respiratory rate is meaningful, but measurement method must be defensible | Cretikos et al., 2008; Subbe et al., 2001 ³³³⁵ |

Implementation checklist for research pipelines

A defensible ECG and PPG respiratory rate pipeline should include seven components.

First, preserve raw waveform access. Respiratory-rate estimates cannot be audited if the source ECG and PPG signals are unavailable.

Second, calculate modality-specific signal quality indices. ECG R-peak quality, ECG morphology quality, PPG pulse quality, and PPG amplitude quality should remain separate.

Third, derive multiple respiratory traces. ECG interval variation, ECG morphology modulation, PPG amplitude variation, PPG intensity variation, and PPG frequency variation provide complementary evidence.¹²

Fourth, reject corrupted windows before fusion. Rejection is part of measurement integrity. A missing value is preferable to a confident artifact.

Fifth, fuse only qualified estimates. Weighted fusion should reflect signal confidence and agreement between traces.²¹²²

Sixth, report uncertainty and missingness. A respiratory-rate value without quality metadata is incomplete.

Seventh, validate in the intended use context. Resting laboratory performance does not guarantee ambulatory performance. Motion, posture, sleep stage, arrhythmia, perfusion, and sensor contact can change error behavior.

This approach fits Sensor Bio's infrastructure framing. The platform should expose signals and quality context so researchers can inspect, reproduce, and improve the pipeline. To discuss biosignal infrastructure for research programs, [request access to Sensor Bio's platform](/get-started/).

Where direct respiratory sensors still matter

ECG and PPG are useful because they are already present in many wearable and bedside biosignal systems. They are not replacements for direct respiratory instrumentation. Respiratory effort studies, pediatric photoplethysmogram studies, and chest clinic studies show that respiratory information can be extracted from PPG, but they also reinforce the need for reference comparison and population-specific validation.^1617181920

A direct respiratory sensor remains the better reference when the question is airflow, ventilation pattern, or respiratory mechanics. ECG and PPG are better understood as scalable cardiovascular proxies for breathing frequency. That framing keeps the metric useful without overstating it. It also aligns with evidence on respiratory rate as a deterioration marker: the signal matters, but measurement method determines whether the number is trustworthy.²⁵³⁹

Methodology also matters for autonomic interpretation. Respiratory sinus arrhythmia is affected by breathing pattern, vagal tone, and quantification choices. HRV norms and respiratory-frequency methods help contextualize the signal, but they do not remove the need for signal inspection.⁴²⁴⁶⁴⁹

What ECG and PPG respiratory rate cannot show alone

ECG and PPG respiratory rate estimates do not measure tidal volume. They do not identify airflow obstruction. They do not diagnose sleep apnea, respiratory infection, or pulmonary disease. They estimate breathing frequency from cardiovascular modulation.

That limitation is not a weakness if the metric is labeled correctly. Breaths per minute can still be useful in longitudinal physiology, sleep research, and remote monitoring workflows when quality rules and validation are explicit. The problem begins when an indirect estimate is presented as a direct respiratory measurement.

The correct standard is methodological transparency. Show the signal. Show the derived traces. Show the quality score. Show what was rejected. Then publish the respiratory rate.

FAQ

Can respiratory rate be measured from ECG?

ECG can estimate respiratory rate indirectly. Respiration modulates the ECG through respiratory sinus arrhythmia, QRS amplitude changes, and heart-axis movement. Algorithms can derive a respiratory trace from RR intervals or ECG morphology, then estimate breaths per minute. The estimate depends on clean R-peak detection, rhythm regularity, electrode contact, and validation against a reference respiratory signal.²⁴⁷

Can respiratory rate be measured from PPG?

PPG can estimate respiratory rate indirectly through respiratory-induced amplitude, intensity, and frequency variation. Breathing changes venous return, stroke volume, vascular tone, and pulse timing. These changes can appear in the optical waveform. PPG-derived respiratory rate requires strong pulse morphology, stable contact, motion-artifact rejection, and agreement across derived respiratory traces.³¹⁰¹⁵

Is ECG or PPG better for wearable respiratory rate?

Neither modality is always better. ECG is often stronger for beat timing when the electrical signal is clean. PPG can carry strong respiratory amplitude modulation, especially when perfusion and contact are stable. The best architecture usually treats ECG and PPG as complementary signals, scores them separately, and fuses only the estimates that pass quality checks.⁴⁵

Why does signal quality matter for respiratory rate from wearables?

Respiratory rate from ECG and PPG is derived from modulation patterns, not direct airflow. Motion artifact, poor contact, low perfusion, baseline wander, ectopic beats, and arrhythmia can all create false respiratory components. Signal quality indices reduce this risk by rejecting corrupted segments before estimating or fusing respiratory rate.⁶⁷²⁶

What reference should wearable respiratory rate be validated against?

Validation should compare ECG- and PPG-derived respiratory rate against a reference respiratory signal. Common reference methods include capnography, respiratory inductance plethysmography, or other accepted respiratory monitoring systems, depending on study design. Validation should report error distributions, rejection rates, missingness, and performance across posture, movement, and physiological state.¹⁴³⁰

Can ECG and PPG respiratory rate detect disease?

ECG and PPG respiratory rate should not be treated as a standalone disease detection method. They estimate breathing frequency from cardiovascular signals. The metric may support research or monitoring workflows when validated for a defined context, but it does not diagnose a condition by itself. Any clinical use requires appropriate validation, governance, and review.