Circadian Rhythm Tracking via PPG: How Pulse Signals Reveal Your Body Clock
Every heartbeat carries a timestamp of your internal body clock, and PPG sensors are learning to read it. The circadian system, the approximately 24-hour biological oscillator centered in the suprachiasmatic nucleus (SCN) of the hypothalamus, exerts pervasive control over cardiovascular function. Heart rate, heart rate variability, blood pressure, vascular tone, and peripheral blood flow all follow predictable daily rhythms that are encoded in the PPG waveform.
The ability to extract circadian information from continuous wrist PPG data has significant implications for health monitoring, sleep medicine, and chronobiology research. Circadian disruption, whether from shift work, jet lag, social jet lag, or circadian rhythm sleep-wake disorders, is associated with increased cardiovascular risk, metabolic dysfunction, and impaired cognitive performance. A wearable sensor that continuously and passively tracks circadian phase could enable early detection of misalignment and personalized timing of interventions. For background on how PPG signals are acquired and processed, see our PPG technology overview.
Circadian Control of Cardiovascular Function
The SCN orchestrates cardiovascular rhythms through two primary pathways: autonomic nervous system modulation and hormonal signaling. Understanding these pathways is essential for interpreting PPG-derived circadian features.
Autonomic Nervous System Rhythms
Sympathetic and parasympathetic tone follow robust circadian patterns. Sympathetic activity, mediated by norepinephrine and epinephrine, increases during the biological day and decreases during the biological night. This pattern drives the circadian variation in heart rate, with rates typically 10-20% higher during waking hours compared to sleep. Parasympathetic (vagal) tone follows the opposite pattern, increasing during sleep and contributing to the nocturnal heart rate nadir.
Scheer et al. (2010; DOI: 10.1073/pnas.0914193107) demonstrated through forced desynchrony protocols that the circadian system independently contributes approximately 5-8 BPM of heart rate variation, even after controlling for sleep-wake state, posture, and physical activity. This endogenous circadian component persists in the absence of behavioral rhythms and represents a direct SCN-driven modulation of cardiac autonomic balance.
The autonomic circadian rhythm is reflected in HRV metrics accessible through PPG. The high-frequency (HF) component of HRV (0.15-0.4 Hz), reflecting parasympathetic modulation, shows a nocturnal peak that is 50-200% higher than daytime values in healthy individuals (Vandewalle et al., 2007). The LF/HF ratio, often interpreted as a marker of sympathovagal balance, follows a circadian pattern with higher values during the day. RMSSD, a time-domain parasympathetic index readily computed from PPG-derived pulse intervals, mirrors the HF power circadian profile.
Vascular Tone and Peripheral Blood Flow
Peripheral vascular resistance follows its own circadian pattern, partially independent of autonomic cardiac effects. Sympathetically mediated vasoconstriction is higher during the biological day, reducing peripheral blood flow and PPG signal amplitude. During the biological night, vasodilation increases peripheral perfusion, producing larger PPG pulse amplitudes.
Kräuchi and Wirz-Justice (1994; DOI: 10.1152/ajpregu.1994.267.3.R819) showed that the distal-proximal skin temperature gradient, which reflects peripheral vasodilation, follows a circadian rhythm closely linked to the propensity for sleep onset. Because PPG amplitude is directly modulated by peripheral blood volume, this vascular rhythm contributes an additional circadian signature to the PPG signal beyond heart rate and HRV.
Cortisol and Blood Pressure Rhythms
The hypothalamic-pituitary-adrenal (HPA) axis produces cortisol with a robust circadian rhythm, peaking 20-30 minutes after awakening (the cortisol awakening response, CAR) and reaching its nadir around midnight. Cortisol influences cardiovascular function through mineralocorticoid and glucocorticoid receptor activation, contributing to the morning surge in blood pressure and heart rate.
Blood pressure follows a circadian pattern with a 10-20% nocturnal dip in normotensive individuals. The absence of this dip (non-dipping pattern) is an independent cardiovascular risk factor. While PPG cannot directly measure blood pressure, pulse wave features related to arterial stiffness and vascular resistance (augmentation index, pulse transit time) reflect these hemodynamic changes and encode circadian blood pressure information indirectly. For more on PPG-derived vascular features, see our article on PPG vascular aging biomarkers.
Extracting Circadian Features from Continuous PPG
Converting continuous wrist PPG data into circadian metrics requires signal processing pipelines that handle the challenges of ambulatory recording: motion artifacts, intermittent sensor contact, and the need to separate circadian from behavioral and ultradian components.
Heart Rate Circadian Profiling
The most straightforward circadian feature from PPG is the 24-hour heart rate profile. After computing beat-to-beat heart rate from PPG pulse intervals, the process involves several steps.
First, motion artifact contaminated segments must be identified and excluded. Accelerometer data, typically co-collected with PPG in wearable devices, enables activity-based filtering. Only periods of low activity (below 50-100 millig threshold) are retained for circadian analysis, as exercise-induced heart rate elevation can mask the underlying circadian signal. For technical details on artifact handling, see our motion artifact removal guide.
Second, the cleaned resting heart rate time series is smoothed to suppress ultradian oscillations (rhythms with periods shorter than 24 hours, such as the 90-minute sleep cycle). Moving average filters, lowpass filters with cutoff frequencies around 0.5-1 cycle/day, or wavelet decomposition can extract the circadian component.
Third, cosinor analysis fits a sinusoidal model to the 24-hour heart rate data: HR(t) = MESOR + Amplitude * cos(2pit/24 - Acrophase), where MESOR is the rhythm-adjusted mean, Amplitude characterizes the strength of the circadian variation, and Acrophase indicates the time of peak heart rate. Typical values in healthy adults are: MESOR of 65-75 BPM, amplitude of 5-10 BPM, and acrophase at 14:00-18:00 local time.
HRV-Based Circadian Markers
HRV metrics provide complementary circadian information that may be more sensitive to circadian phase than heart rate alone. RMSSD, computed from consecutive PPG pulse intervals during rest periods, is particularly useful because it selectively reflects parasympathetic tone, which is under strong circadian control.
de Zambotti et al. (2016; DOI: 10.1016/j.smrv.2015.01.006) demonstrated that wrist PPG-derived HRV indices tracked laboratory polysomnographic HRV with correlations of r = 0.75-0.85 for RMSSD during sleep. The circadian profile of RMSSD typically shows a peak during the first half of the sleep period, declining toward morning in parallel with the cortisol awakening response.
Spectral HRV analysis adds frequency-domain circadian features. The LF/HF ratio follows a circadian pattern with values of 1.0-2.0 during daytime and 0.5-1.0 during nighttime in healthy individuals. However, PPG-derived spectral HRV requires careful handling of interpolation artifacts and ectopic beat correction, as these can introduce spurious spectral components that distort circadian profiles.
Pulse Wave Amplitude and Peripheral Vascular Indices
PPG pulse wave amplitude (PWA), defined as the peak-to-trough amplitude of the AC component normalized by the DC component, varies with peripheral vasomotor tone and follows a circadian pattern. PWA is typically higher during the biological night (vasodilation) and lower during the biological day (vasoconstriction).
Averaging PWA over 5-10 minute windows and fitting cosinor models yields circadian amplitude and phase estimates that complement heart rate-derived measures. The advantage of PWA is its sensitivity to peripheral vascular tone changes that are under direct sympathetic circadian control, providing information somewhat independent of the cardiac autonomic balance reflected in heart rate and HRV.
Circadian Phase Estimation Algorithms
The ultimate goal of PPG circadian tracking is estimating the phase of the internal circadian clock, traditionally assessed by dim-light melatonin onset (DLMO), a laboratory procedure requiring serial saliva or blood samples in controlled lighting conditions.
Single-Feature Cosinor Methods
The simplest phase estimation approach fits a single cosinor model to 24-48 hours of resting heart rate data. The fitted acrophase is then used to estimate DLMO by applying a population-average offset (heart rate acrophase typically leads DLMO by approximately 7-9 hours).
Woelders et al. (2017; DOI: 10.1177/0748730417741687) evaluated this approach in 16 healthy volunteers with simultaneous DLMO assessment and ambulatory heart rate monitoring. Mean phase estimation error was 1.8 hours (SD 1.2 hours), with errors attributable primarily to individual variation in the heart rate-to-DLMO phase relationship.
Multi-Feature Machine Learning Approaches
Combining multiple PPG-derived features with activity and temperature data improves circadian phase estimation. Stone et al. (2019; DOI: 10.1177/0748730419885171) developed a Bayesian state-space model integrating heart rate, activity, and skin temperature to estimate circadian phase from wearable data. Their model achieved a mean absolute error of 1.0 hour compared to DLMO in a laboratory study of 34 participants.
Cheng et al. (2021; DOI: 10.1073/pnas.2104985118) used a convolutional neural network trained on wrist accelerometry and heart rate data from 214 participants with laboratory-assessed circadian phase. Their model, named TimeShift, achieved a mean absolute DLMO prediction error of 0.83 hours in a validation cohort, approaching the test-retest reliability of DLMO itself (approximately 0.5-1.0 hours).
These results demonstrate that PPG-based circadian phase estimation is approaching clinical utility for identifying circadian misalignment, though the accuracy gap with laboratory methods remains significant for applications requiring precise phase information (such as optimally timed light therapy or melatonin administration).
Applications in Health and Disease
Shift Work and Jet Lag Assessment
Shift work affects approximately 15-20% of the workforce in industrialized countries and is associated with increased risks of cardiovascular disease (relative risk 1.17-1.40), metabolic syndrome, and cancer. PPG-based circadian monitoring could enable real-time assessment of circadian adaptation during shift rotations, identifying workers whose internal clocks have not adjusted and who are at highest risk for adverse health outcomes and workplace errors.
Bowman et al. (2021) demonstrated that wearable-derived circadian features (heart rate amplitude, activity rhythm stability, and interdaily stability) could classify shift workers as adapted or non-adapted with 78% accuracy, compared to DLMO-based classification. This accuracy, while imperfect, is sufficient for population-level screening and could be improved with personalized baseline models.
Sleep-Wake Disorder Characterization
Circadian rhythm sleep-wake disorders (CRSWDs), including delayed sleep-wake phase disorder (DSWPD), advanced sleep-wake phase disorder (ASWPD), and non-24-hour sleep-wake rhythm disorder, are characterized by misalignment between the internal circadian clock and the desired sleep-wake schedule. Diagnosis currently relies on sleep diaries and actigraphy over 7-14 days, supplemented by DLMO when available.
Continuous PPG monitoring could enhance CRSWD assessment by providing circadian phase estimates alongside sleep detection, enabling clinicians to quantify the degree of phase misalignment directly. The combination of PPG-derived sleep staging (using heart rate and HRV patterns) with circadian phase estimation represents a potential single-sensor solution for comprehensive sleep-circadian assessment.
Chronotherapy Optimization
The efficacy of many medications varies with circadian timing. Antihypertensive drugs, chemotherapy agents, and anti-asthma medications all show time-of-day-dependent efficacy and toxicity profiles. PPG-based circadian phase estimation could enable personalized chronotherapy, timing medication administration to the individual's internal biological time rather than clock time.
This application requires phase estimation accuracy better than 2 hours to be clinically meaningful, a threshold that current multi-feature algorithms are approaching but have not consistently achieved in free-living conditions.
Technical Challenges
Masking Effects
The greatest challenge in PPG-based circadian analysis is separating endogenous circadian signals from exogenous masking effects. Physical activity, posture changes, emotional stress, caffeine, alcohol, and ambient temperature all affect heart rate, HRV, and peripheral vascular tone. These behavioral and environmental influences can mask or distort the underlying circadian signal.
Activity-based filtering (analyzing only rest periods) mitigates this problem but does not eliminate it, as even at rest, postural differences between sitting and lying affect cardiovascular parameters. Advanced algorithms attempt to model and remove masking effects explicitly, but the accuracy of this demasking process is fundamentally limited by the incompleteness of the available contextual data.
Individual Variability
The phase relationship between PPG-derived circadian markers and the central pacemaker (DLMO) varies substantially across individuals. Heart rate acrophase relative to DLMO ranges from 5 to 11 hours across published studies. This variability reflects genuine physiological differences in the coupling between the SCN and downstream effector systems, and it limits the accuracy of any algorithm that applies population-average phase offsets.
Personalized models that learn an individual's characteristic phase relationships over time could substantially improve accuracy but require extended baseline data collection and periodic recalibration against ground-truth phase markers.
Data Quality in Free-Living Conditions
Continuous wrist PPG in free-living conditions produces variable-quality data. Sensor displacement during sleep, motion during daily activities, and environmental light interference create gaps and artifacts in the heart rate and HRV time series. Circadian analysis algorithms must be robust to these data quality issues, either through gap-tolerant fitting methods or through adaptive quality weighting schemes.
For researchers and engineers developing circadian analysis pipelines, our algorithms reference provides foundational guidance on PPG signal processing, and our overview of PPG in health conditions contextualizes the clinical significance of circadian disruption.
Conclusion
PPG-based circadian rhythm tracking is transitioning from a research curiosity to a practical capability. The cardiovascular system's robust circadian rhythmicity means that continuous PPG recordings inherently contain information about the phase, amplitude, and regularity of the internal body clock. Current algorithms can estimate circadian phase with accuracy approaching 1 hour under controlled conditions, and ongoing advances in machine learning and multi-sensor fusion are narrowing the gap with laboratory gold standards.
The clinical and public health implications are substantial. Circadian disruption is increasingly recognized as a modifiable risk factor for cardiovascular disease, metabolic disorders, and mental health conditions. A widely deployed, passive circadian monitoring capability through existing wearable PPG sensors could enable population-scale screening for circadian misalignment, personalized shift work scheduling, and optimized timing of medical interventions. The scientific foundation is strong; the engineering challenge now is achieving reliable performance in the noisy, variable conditions of real-world wearable use.