Locomotor-Autonomic Coupling is Phase-Lagged and Behaviorally Driven: Evidence from long-term simultaneous heartrate and activity monitoring

Locomotor-Autonomic Coupling is Phase-Lagged and Behaviorally Driven: Evidence from long-term simultaneous heartrate and activity monitoring | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Locomotor-Autonomic Coupling is Phase-Lagged and Behaviorally Driven: Evidence from long-term simultaneous heartrate and activity monitoring Sanaz Ghaffari, Olivier Demers, Masoumeh Goudarzi, Abass Zakari, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7108155/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 26 Sep, 2025 Read the published version in Scientific Reports → Version 1 posted 10 You are reading this latest preprint version Abstract The temporal coordination between behavioral and autonomic rhythms is a critical feature of circadian physiology, yet the precise alignment and causal structure of this coupling remain poorly characterized in free-living humans. Using long-term wearable data spanning 30 days from 52 individuals, we analyzed accelerometry (ACC) and heart rate (BPM, via IBI) to quantify circadian phase alignment, inter-day stability, and temporal directionality between locomotor and autonomic systems. Across individuals, behavioral activity rhythms consistently peaked earlier than autonomic rhythms (mean lag: –1.8h, p < 0.001), with the lag largely attributable to greater variability in locomotor phase. Despite this temporal dissociation, both signals exhibited coherent 24-hour patterns and relatively stable inter-day acrophases. Lag magnitude was negatively correlated with nighttime BPM (r = –0.55, p < 0.001), suggesting a link between autonomic hyperactivation and desynchrony. Crucially, behavioral acrophase more strongly predicted daily lag fluctuations than BPM acrophase, and causal analyses revealed asymmetric dependencies: same-day activity levels were significantly predictive of nighttime heart rate, whereas the reverse relationship was weaker and less consistent. Granger causality confirmed a predominant flow from ACC to BPM across subjects ( p = 0.0045). These findings establish that autonomic rhythms lag behind and are shaped by preceding behavioral activation, supporting a behavior-first model of internal circadian organization. Biological sciences/Neuroscience Biological sciences/Physiology Biological sciences/Psychology Social science/Psychology Figures Figure 1 Figure 2 Figure 3 Introduction Circadian rhythms govern a wide range of physiological and behavioral processes, including sleep–wake cycles, core body temperature, hormone secretion, locomotor activity, and autonomic function. In humans, these rhythms are coordinated by a central pacemaker in the suprachiasmatic nucleus (SCN), yet they are expressed across multiple organ systems with varying degrees of independence and synchrony. Recent advances in wearable sensing technology have enabled large-scale, longitudinal monitoring of circadian patterns in free-living conditions, offering unprecedented resolution into human rhythms outside the laboratory. Several recent studies have leveraged wearable data to quantify circadian physiology at population scale. For instance, Natarajan et al. [1] analyzed over 19,000 individuals using Fitbit data and identified consistent 24-hour rhythms in heart rate and activity, revealing systematic phase differences between these signals and strong modulation by age and sex. Similarly, Shim et al. [2] demonstrated that circadian parameters derived from accelerometry are predictive of aging and healthspan in a large cohort, positioning these digital rhythms as biomarkers of physiological resilience. In a related effort, Huang et al. [3] introduced a computationally efficient method to estimate circadian phase from wearable heart rate and temperature data, validating its accuracy across both lab-controlled and real-world datasets. Morin et al. [4] have emphasized the importance of behavioral-physiological alignment, framing circadian misalignment as a key contributor to adverse outcomes in sleep, metabolism, and mental health. However, the degree to which autonomic rhythms—particularly heart rate—are temporally coupled with behavioral signals such as movement remains incompletely understood. Prior work by Karimi et al. [5] used smartwatch data to characterize the longitudinal dynamics of heart rate and activity, showing rich variation across individuals and time, yet without explicitly quantifying their phase relationships. Similarly, Hallgrimsson et al. [6] developed machine learning models to learn personalized cardiovascular responses from wearable sensor data, but focused on predictive performance rather than circadian structure. Emerging evidence suggests that commercial wearables can detect and compare multi-system rhythms—including activity, heart rate, and core body temperature—in home environments [7]. Yet, these studies often lack the temporal resolution and duration required to assess causal directionality or inter-system coordination. Leota et al. [8], using large-scale actigraphy data, found that sleep timing significantly influences next-day activity levels, highlighting the importance of temporal sequencing in behavioral regulation. Building on this, Kim et al. [9] extended phase modeling techniques to both heart rate and body temperature across more than 50,000 participant-days, confirming circadian structure but leaving open questions about inter-system alignment. Meanwhile, sleep detection via heart rate alone has been validated in naturalistic settings [10], underscoring the feasibility of deriving autonomic states from wearable data. Together, these findings highlight the need to move beyond descriptive analyses and toward a mechanistic understanding of how autonomic and behavioral rhythms interact over time. Specifically, it remains unresolved whether the autonomic system merely reflects behavioral patterns or exhibits independent, potentially leading dynamics. Resolving this question requires high-frequency, long-duration, multi-modal recordings capable of detecting both phase relationships and causal flow between physiological systems under real-world conditions. In this study, we leverage 30 days of high-resolution, simultaneous accelerometry and heart rate recordings from 52 individuals to examine the temporal structure, alignment, and causal relationship between behavioral and autonomic circadian rhythms in naturalistic conditions. By quantifying acrophase timing, day-to-day stability, and lag between systems, we test whether the autonomic system lags behind behavioral output, whether this lag varies across individuals, and whether the direction of influence is consistently behavior-first. Our goal is to resolve whether the heart follows the body—or leads it—in daily life, and to uncover the individual and physiological factors that shape this internal coordination. Materials and Methods Fifty-two healthy young adults (26 male, 26 female; age range: 18–38 years) were recruited from the Bishop’s University and Université de Sherbrooke communities. All participants provided written informed consent in accordance with ethics approval from the Centre intégré universitaire de santé et de services sociaux de l’Estrie – Centre hospitalier universitaire de Sherbrooke (CIUSSS de l’Estrie – CHUS) . All methods were performed in accordance with the relevant guidelines and regulations. Participants were screened to exclude any neurological or psychiatric disorders, sleep disorders, or use of medications affecting the central nervous system. Smartwatch Data Acquisition Each participant wore a Samsung Galaxy Watch Active2 continuously over a 30-day monitoring period. This device features: 8-channel green LED photoplethysmography (PPG) sensor array 3-axis accelerometer and gyroscope 360 x 360 Super AMOLED display (1.4") Tizen OS (programmable environment with direct sensor access) Battery capacity: 340 mAh (approximately 36 hours per charge) Custom firmware written in C was deployed on the watches to enable continuous data sampling in energy-efficient bursts. Specifically, the accelerometer and PPG sensors were activated for 1 minute every 10 minutes , recording at 10 Hz , resulting in 4,320 minutes of data per subject. This sparse sampling protocol allowed high-frequency physiological capture with minimal battery strain. Participants were instructed to wear the device at all times, including overnight, except during charging. Charging was typically done once daily during low-activity windows and took under 30 minutes. Subjects were encouraged to prioritize nighttime wear due to the reduced motion artifact and superior PPG signal quality during sleep. Adherence was verified visually during postprocessing. Preprocessing and Hourly Matrix Construction Raw smartwatch sensor data were processed to extract hourly-resolved time series of accelerometry and interbeat interval (IBI) values for each subject. For each of the 52 participants, all available 10 Hz sensor data (accelerometer and photoplethysmography [PPG]) were loaded from local CSV files and segmented into non-overlapping 10-second epochs. Accelerometry: For each epoch, the Euclidean norm of the 3-axis accelerometer signal was computed to yield a scalar magnitude of movement. These values were binned by local clock time into a 24 (hour) × n (day) matrix, where each cell represents the mean accelerometer magnitude for that hour and day. IBI Estimation: The raw PPG signal was bandpass-filtered (0.35–4.0 Hz) and resampled to 30 Hz to improve peak detection fidelity. Epochs were accepted for IBI estimation only if they passed a signal quality check based on normalized autocorrelation, ensuring the presence of physiologically plausible periodicity. For accepted epochs, peaks were detected using amplitude thresholding and minimum peak distance criteria. Instantaneous IBI values (in seconds) were calculated as the mean time between adjacent peaks. These were also aggregated into a 24 × n matrix per subject, where each cell represents the average IBI for a given hour and day. IBI were converted to BPM for further analysis. Time Alignment and Output: Timestamps for all sensor data were converted to Eastern Time (EST/EDT) using timezone-aware conversion to ensure consistent alignment with local clock hour. Because both accelerometry and IBI were measured concurrently using the same device, no temporal alignment between behavioral and autonomic data streams (same timestamps applied to both signals). Only valid epochs with usable sensor data were included. Each subject’s data were stored as a dictionary containing a list of dates and two matrices (acc, ibi), each with shape (24 × n_days ). All subjects’ matrices were saved into a single pickle file for downstream analysis . Rhythm Analysis, Modeling, and Statistical Inference Circadian Feature Extraction Following preprocessing and alignment of the 24×N (hour × day) matrices for accelerometry (ACC) and inter-beat interval (IBI), we conducted subject-level and population-level analyses to quantify circadian structure, acrophase timing, and inter-system alignment. For each subject, mean hourly ACC and IBI values were computed across valid days, and IBI was converted to beats per minute (BPM) as BPM = 60 / IBI. To characterize 24-hour rhythms, we fit a fundamental cosinor model to each signal: y(t) = β₀ + β₁ · cos(2πt / 24) + β₂ · sin(2πt / 24) where: β₀ is the mesor (mean level), β₁, β₂ are the cosine and sine amplitudes. using ordinary least squares. Acrophase was derived from the fitted coefficients as: ϕ = (atan2(β₂, β₁) mod 2π) · (24 / 2π) Phases occurring before 06:00 were wrapped forward by 24 hours to enable consistent comparison across subjects and systems. Per-subject cosinor fits were overlaid with raw 24-hour mean signals and visually inspected to identify poor fits or sparse coverage (N=6 subjects excluded from downstream analyses). Group-Level Comparisons To assess sex and age differences in circadian profiles, subjects were grouped by sex and age (<30 vs. ≥30 years), and group-wise grand mean ± standard error curves were plotted. At each hour, unpaired t -tests were used to compare group means, with statistically significant hours annotated. Differences in acrophase timing across groups (sex, age) and modalities (ACC vs. BPM) were evaluated using either paired or unpaired t -tests, as appropriate. In all tests, normality assumptions were visually verified. Effect sizes (mean differences in hours) and test statistics (t, p) were reported. Circadian Coupling Metrics To examine coupling between locomotor (ACC) and autonomic (BPM) rhythms, we computed daily acrophases for each signal and calculated per-day lag as: Δϕ_lag = ((ϕ_ACC - ϕ_BPM + 12) mod 24) – 12 Subject-level lag distributions were compared across sex and correlated with physiological features, including nighttime BPM (1–5 AM), 24-h BPM standard deviation, and locomotor rhythm consistency (mean pairwise inter-day Pearson r ). ACC consistency was computed from the lower triangle of inter-day correlation matrices derived from the daily ACC profiles. Variability and Predictive Models To assess variability in acrophase timing across days, we computed the standard deviation of daily acrophases per subject for each signal. Paired t -tests were used to compare across signals. Additionally, for each subject, we fit linear models predicting daily lag from either ACC or BPM acrophase, computing R² as the proportion of variance explained. Paired t -tests across subjects compared R² values for the two predictors. Temporal Asymmetry and Causal Directionality We investigated temporal asymmetry in autonomic–behavioral coupling via two approaches: Day-to-night correlation asymmetry : For each subject, we correlated: Previous-day daytime ACC (9–17h) with same-night BPM (1–4h) Nighttime BPM with next-day ACC Mean correlations were compared using a paired t -test. Granger causality : We tested for unidirectional influence (lag=1) between daytime ACC and nighttime BPM using the grangercausalitytests function from statsmodels. Two models were tested per subject: ACC[t-1] → BPM[t] BPM[t-1] → ACC[t] Distributions of p-values were compared using the Wilcoxon signed-rank test to assess group-level asymmetry. Visualization and Summary Statistics All data visualization was performed using matplotlib, seaborn, and numpy. Histogram bins, violin plots, scatter plots with regression fits, and subject-level heatmaps were used to illustrate rhythmic profiles, acrophase variability, inter-system lag, and consistency metrics. All statistical analyses were performed in Python using: scipy.stats for t -tests, Pearson correlations, standard deviations statsmodels for Granger causality sklearn.linear_model.LinearRegression for acrophase-lag R² computation circmean and circstd from scipy.stats for circular statistics Statistical significance was assessed at α = 0.05 unless otherwise specified, and p -values were corrected where relevant (e.g., per-hour testing) through visual annotation but not formal multiple-comparisons correction. Results Figure 1 illustrates the temporal structure, group-level dynamics, and individual variability in circadian rhythms derived from accelerometry (ACC) and heart rate (BPM) data in our full cohort (n = 52). Panel A shows an unraveled, 44-day time series of z-scored ACC and IBI (inverse of BPM) in a single subject (S36) revealing coherent circadian oscillations with substantial inter-day modulation. ACC (blue) and IBI (orange) exhibit tight anti-correlated oscillations, with activity peaking during day (white background) and IBI peaking during night (gray background). Panel B1 displays grand mean 24-h rhythms, highlighting a robust diurnal pattern for both signals, peaking in the late afternoon. Sex differences are shown in B2–B3, with females exhibiting significantly elevated early-morning BPM compared to males (* p < 0.05), while ACC showed a modest trend but did not reach significance. Age differences (Panels B4–B5) revealed lower activity and BPM in older adults (≥ 30 years) compared to younger adults (< 30), particularly in the early afternoon, but age differences were not significant in this cohort. Bar plots in Panels C1–C5 quantify acrophase differences across signals and subgroups. ACC rhythms peaked significantly earlier than BPM rhythms, with a mean phase difference of 1.78 hours (C1; t = 2.80, p = 0.007), indicating a temporal lag between behavioral activation and subsequent autonomic arousal. Males exhibited a significantly later acrophase than females in ACC signal (C2; Δ = 2.20h, t = 2.12, p = 0.040), but not BPM (C3; p > 0.2). No significant age-related differences in acrophase were observed for either ACC (C4; p = 0.948) or BPM (C5; p = 0.966). Together, these findings suggest a modest but reliable lag between motor and cardiac rhythms, a sex difference in circadian timing, and stable phase alignment across age groups. Panel D presents raw individual-level 24-hour profiles of accelerometry (ACC, blue) and heart rate (BPM, red) across all 52 participants. Solid lines depict the mean time series, while dashed lines show fitted cosinor curves with corresponding acrophase markers (dots). Most participants exhibited coherent circadian patterns in both signals, with moderate alignment in acrophase and waveform morphology. However, six participants (highlighted in red: s14, s31, s42, s8, s3, s18) demonstrated insufficient data coverage or unreliable cosinor fits and were therefore excluded from further acrophase and rhythm strength analyses. This visual overview underscores substantial inter-individual variability in rhythm amplitude, phase, and coherence across behavioral and autonomic domains. Figure 2 investigates the inter-individual variability and determinants of lag between behavioral (ACC) and autonomic (BPM) rhythms. Panel A shows the distribution of ACC–BPM acrophase lag across male/female split, males and females showed overlapping lag distributions and females showing a slightly longer lag (non-significant, p = 0.276). Panel B examines the hour-by-hour inter-subject pearson correlation between hourly BPM and lag, revealing a sinusoidal correlation pattern (subjects with greater physiological lag (more negative) had higher BPM at night). Panel C highlights the heterogeneity in locomotor circadian stability using two exemplar participants—one with the highest inter-day consistency and one with among the lowest. The left plots show daily ACC heatmaps by hour, and the right-side plots show the corresponding inter-day ACC correlation matrices. High-consistency subjects show clear diurnal structure and high inter-day correlation, whereas low-consistency individuals show noisy or inconsistent patterns. Regression analyses in panels D–G explore predictors of ACC–BPM lag. A robust negative correlation was observed between nighttime BPM and lag (r = − 0.55, p < 0.001), indicating that higher nocturnal BPM was associated to poorer autonomic/locomotor alignment. However, inter-day ACC consistency (E) and BPM variability (G) were not significantly associated with lag. Panels F–H assess the synchrony and variability of ACC and BPM acrophases. Individual ACC and BPM acrophases were not correlated across subjects (F; r = 0.02, p = 0.901), and the lag between autonomic and locomotor activity was due almost entirely to variability in locomotor activity (r = 0.84, p < 0.001), confirming that the lag reflects a shifted locomotor rhythm rather than shifted autonomic response. Supporting this, panel I demonstrates that ACC acrophase variability across days was significantly greater than BPM variability, reflecting stronger rhythm regularity in the autonomic system. Figure 3 provides converging evidence that behavioral activity precedes and potentially regulates autonomic rhythms, based on pooled and intra-subject analyses of circadian acrophase and causal inference. Panel A shows the overall distribution of acrophase values for ACC and BPM (via IBI), pooled across all days and subjects. ACC rhythms peak significantly earlier than BPM (mean = 14.6h vs. 16.8h, t = − 14.90, p < 0.0001), confirming a consistent behavioral lead. Despite this phase offset, the inter-day variability of acrophase (Panel B) was comparable between signals (mean SD = 1.96h for ACC vs. 1.69h for BPM, t = 1.28, p = 0.209), indicating that both systems exhibit relatively stable phase timing across days within subjects. When pooled across all days and subjects, higher nighttime BPM (1–4 AM) was associated with a more negative ACC–BPM lag (Panel C, r = − 0.22, p < 0.0001), consistent with prior findings that elevated nocturnal heart rate reflects a longer lag between activity and subsequent BPM changes. Panel D compares within-subject predictive power (r²) of daily acrophase values for explaining daily ACC–BPM lag within each subject. ACC acrophase more strongly predicted the magnitude and direction of the daily lag than BPM acrophase, though the difference approached but did not reach significance (t = 1.75, p = 0.090). To probe temporal directionality, panels E and F evaluate causal asymmetry in the relationship between behavior and autonomic output. Panel E shows that daily activity levels more strongly correlate with that-night BPM (ACC→BPM) than the reverse (nightly BPM correlating with next-day acc, BPM→ACC), with a significant paired t-test (p = 0.0039). Granger causality analysis (Panel F) supports this interpretation: a larger number of participants exhibited significant unidirectional Granger-causality from daytime ACC to that-night BPM (yellow bars at low p-values), with a Wilcoxon test confirming this asymmetry (p = 0.0045). Together, these findings suggest a behavior-first model of internal coupling, where daytime activity shapes subsequent autonomic regulation rather than being driven by it. Discussion Our 30-day, high temporal resolution investigation of locomotor and autonomic rhythms provides compelling evidence that behavioral activity reliably precedes—and likely shapes—autonomic patterns, challenging the conventional notion that a central pacemaker independently governs both systems. We observed a robust phase lag (mean ≈ − 1.8 h) wherein activity acrophase precedes heart rate acrophase. This lead was consistent across individuals and largely unaffected by age or sex. This behavioral lead is aligned with individual-level findings showing that daily behaviors such as physical activity, diet, and stress can influence nocturnal heart rate variability and amplitude [ 11 ]. Moreover, our Granger causality (GC) analyses support this directional relationship, reinforcing earlier demonstrations of bidirectional, sleep-stage–dependent brain–heart coupling using EEG and ECG data [ 12 ]. These GC results are further validated by prior methodological reviews that highlight the use of GC in modeling predictive flow between interconnected physiological systems, including cardiovascular and respiratory dynamics [ 15 ]. While both activity and heart rate exhibited strong 24-hour rhythmicity, we found that the inter-individual stability of autonomic acrophase was significantly higher than that of locomotor rhythms. This asymmetry supports the hypothesis that autonomic outputs are more tightly constrained by internal timing mechanisms—potentially including intrinsic cardiac oscillators—than behavioral activity, which may be more influenced by social and environmental cues. This finding complements earlier work demonstrating that heart rate rhythms derived from wearable devices show greater consistency than other physiological indicators across days [ 13 ]. Further supporting the functional impact of desynchrony, we found that higher nighttime heart rate was associated with more negative lag values—i.e., greater misalignment between activity and autonomic rhythms. This pattern replicates previous findings linking elevated nocturnal BPM with circadian disruption and autonomic hyperarousal [ 11 ], and supports its use as a potential digital biomarker for circadian misalignment in real-world populations. Large-scale studies have shown that circadian disruption, as inferred from heart rate patterns, is associated with elevated risk for mood disorders and poor mental health outcomes [ 17 ]. Our study benefits from high-density, 30-day continuous wearable recordings, overcoming the temporal limitations of traditional laboratory studies. While many prior studies using wearables have focused on descriptive analyses or machine learning prediction, our integration of causal modeling places this work in the growing field of chronophysiological inference. This aligns with the broader aims of chronomedicine, where wearable-derived circadian metrics are now being explored for diagnostic, therapeutic, and monitoring applications [ 14 ]. Still, several limitations deserve mention. First, while Granger causality captures directional predictive relationships, it does not establish mechanistic causation. Confounding variables—such as light exposure, psychological state, or caffeine intake—may jointly affect both behavior and autonomic outputs. Future research should integrate wearable data with contextual sensors and experimental manipulations (e.g., scheduled activity paradigms or forced desynchrony protocols) to test whether behavioral changes can shift or realign autonomic rhythms [ 14 ]. Second, our sample size (n = 52), while rich in temporal resolution, is modest compared to population-scale datasets. Nonetheless, the within-subject nature of our analyses supports strong inferences about temporal coordination. Future studies in larger, more diverse cohorts will allow for exploration of demographic and clinical moderators of behavioral-autonomic coupling, including chronotype, shift work, and sleep disorders. Finally, although our study focused on accelerometry and heart rate, a more complete understanding of circadian regulation will require integration of additional physiological signals—including heart rate variability, core body temperature, hormonal profiles, and sleep staging—many of which can now be noninvasively captured using modern wearables [ 13 ]. A systems-level perspective, grounded in temporal modeling approaches [ 15 ], will be essential for mapping the multidimensional landscape of human circadian physiology. Conclusion Our findings support a model in which daily behavioral activity drives and entrains autonomic rhythms. This opens the door for behaviorally timed interventions—such as scheduled exercise or meal timing—to realign internal rhythms, especially in populations vulnerable to circadian disruption. Furthermore, nighttime heart rate emerges as a simple, objective, and real-time biomarker of internal misalignment, detectable with inexpensive consumer-grade devices. Long-term wearable data in naturalistic conditions reveal a consistent pattern: the body leads, and the heart follows. These findings refine our understanding of inter-system circadian organization and provide an empirical basis for using behavior to manage internal synchrony. Declarations Acknowledgements: this work was funded by the National Science and Engineering Research Council of Canada (NSERC) Discovery Grant program RGPIN1507. Author Contributions: S.G. recruited participants and led data acquisition, including MRI and smartwatch recordings. O.D. developed the smartwatch firmware in C for raw data capture and extraction. MG, AZ, and CL performed statistical analyses and signal processing. R.B. conceived and supervised the study, provided funding, and oversaw all aspects of study design and interpretation. All authors contributed to manuscript preparation and approved the final version. Competing Interests: the authors declare no conflict of interest. Data Availability: All raw and preprocessed data generated and analyzed during this study are available from the corresponding author ( [email protected] ) upon reasonable request. Ethics declarations and approval for human experiments: All participants provided written informed consent in accordance with ethics approval from the Centre intégré universitaire de santé et de services sociaux de l’Estrie – Centre hospitalier universitaire de Sherbrooke (CIUSSS de l’Estrie – CHUS) . Funding: this work was funded by the National Science and Engineering Research Council of Canada (NSERC) Discovery Grant program RGPIN1507. References Natarajan, A., Gleichauf, K., Khalid, M., Heneghan, C., & Schneider, L. D. (2025). Circadian rhythm of heart rate and activity: a cross-sectional study. Chronobiology Int. 42 , 108–121. Shim, J., Fleisch, E., & Barata, F. (2024). Circadian rhythm analysis using wearable-based accelerometry as a digital biomarker of aging and healthspan. npj Digit. Med. 7 , 146. Huang, Q., Patel, N., & Forger, D. B. (2023). Efficient assessment of real-world dynamics of circadian rhythms in free-living conditions. J. R. Soc. Interface 20 , 20230030. Morin, C. M., Carrier, J., Bastien, C., & Harvey, A. G. (2025). Circadian rhythms revealed: unraveling the genetic, physiological, and behavioral tapestry of the human biological clock and rhythms. Frontiers in Sleep. 1 , 1544945. Karimi, F., Amoozgar, Z., Reiazi, R., Hosseinzadeh, M., & Rawassizadeh, R. (2022). Longitudinal analysis of heart rate and physical activity collected from smartwatches. arXiv , abs/2211.08628. Hallgrimsson, H. T., Jankovic, F., Althoff, T., & Foschini, L. (2018). Learning individualized cardiovascular responses from large-scale wearable sensor data. arXiv , abs/1812.01696. Wu, F., Langer, P., Shim, J., Fleisch, E., & Barata, F. (2024). Comparative efficacy of commercial wearables for circadian rhythm home monitoring from activity, heart rate, and core body temperature. arXiv , abs/2404.03408. Leota, J., Czeisler, C. A., Le, F., et al. (2025). Sleep duration and timing are associated with next-day physical activity: insights from two large-scale wearable sensor studies. Proc. Natl. Acad. Sci. U. S. A. 122 , e2420846122. Kim, D., Mayer, C., Lee, M. P., Choi, S. W., Tewari, M., & Forger, D. B. (2023). Efficient assessment of real-world dynamics of circadian rhythms in heart rate and body temperature from wearable data. J. R. Soc. Interface 20 , 20230030. Perez-Pozuelo, I., et al. (2022). Detecting sleep outside the clinic using wearable heart rate devices. Sci. Rep. 12 , 7956. Matias, M., et al. (2022). What possibly affects nighttime heart rate? Conclusions from N-of-1 observational data. Digit. Health . Bashan, A., et al. (2022). Brain and brain–heart Granger causality during wakefulness and sleep. Front. Neurosci . Taguchi, S., et al. (2021). A method for characterizing daily physiology from widely used wearable devices, including HR circadian rhythms. Patterns 2 , 100332. Arnaldi, D., et al. (2024). Wearables in chronomedicine and interpretation of circadian health. Diagnostics (Basel) 15 , 327. Pichot, V., Corbier, C., & Chouchou, F. (2024). The contribution of Granger causality analysis to our understanding of cardiovascular homeostasis: from cardiovascular and respiratory interactions to central autonomic network control. Front. Netw. Physiol. 2 , 1315316. Durgin, J., & Heath, E. (2023). The cardiac circadian clock: implications for cardiovascular health. JACC Basic Transl. Sci. 8 , 1–13. Lee, M. P., Kim, D. W., Fang, Y., Kim, R., Bohnert, A. S. B., Sen, S., & Forger, D. B. (2024). The real-world association between digital markers of circadian disruption and mental health risks. npj Digit. Med. 7 , 13. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 26 Sep, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 05 Aug, 2025 Reviews received at journal 03 Aug, 2025 Reviewers agreed at journal 29 Jul, 2025 Reviews received at journal 27 Jul, 2025 Reviewers agreed at journal 17 Jul, 2025 Reviewers invited by journal 17 Jul, 2025 Editor assigned by journal 17 Jul, 2025 Editor invited by journal 17 Jul, 2025 Submission checks completed at journal 15 Jul, 2025 First submitted to journal 15 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7108155","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":488743324,"identity":"c92d2cc0-4794-4e94-a108-10fd821d23dc","order_by":0,"name":"Sanaz Ghaffari","email":"","orcid":"","institution":"Bishop's University","correspondingAuthor":false,"prefix":"","firstName":"Sanaz","middleName":"","lastName":"Ghaffari","suffix":""},{"id":488743325,"identity":"05aa1ff9-d505-4d2a-bdd3-017902345ebc","order_by":1,"name":"Olivier Demers","email":"","orcid":"","institution":"Université de Sherbrooke","correspondingAuthor":false,"prefix":"","firstName":"Olivier","middleName":"","lastName":"Demers","suffix":""},{"id":488743326,"identity":"a19d91ed-969e-4680-9120-64474d9b34a7","order_by":2,"name":"Masoumeh Goudarzi","email":"","orcid":"","institution":"Bishop's University","correspondingAuthor":false,"prefix":"","firstName":"Masoumeh","middleName":"","lastName":"Goudarzi","suffix":""},{"id":488743327,"identity":"bf66164f-0600-4e5b-8baf-426f872a2a02","order_by":3,"name":"Abass Zakari","email":"","orcid":"","institution":"Bishop's University","correspondingAuthor":false,"prefix":"","firstName":"Abass","middleName":"","lastName":"Zakari","suffix":""},{"id":488743328,"identity":"e905b5d2-3ad8-4b1f-b4c6-4080b816ac84","order_by":4,"name":"Chen Li","email":"","orcid":"","institution":"Bishop's University","correspondingAuthor":false,"prefix":"","firstName":"Chen","middleName":"","lastName":"Li","suffix":""},{"id":488743329,"identity":"36be2c16-b7f9-4004-8d68-87c4b708e0f9","order_by":5,"name":"Russell Butler","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1klEQVRIiWNgGAWjYDACCSB+wMAgxyDB2MBMrBbGhgQGBmPStSQ2AK0jTov57B7zBwkV99I33G5u/lxQcceegf3wA7xaZO6cMWxIOFOcu+HOwTbpGWeeJTbwpBngd5dEjmFDYltC7oYbiW3MvG2HE4BOJUbLv4R0gxuJzZ+BWuwZJNg/EKGlISEBqKVBGqiFsUGCh4AtMscKZyQcSzCcCfILD9AvbTw5Bfi1SDdv+PChJkGe73b74888wBDjZz++Aa8WdHCAgY0k9WAto2AUjIJRMArQAQCBbUpK/IuUOQAAAABJRU5ErkJggg==","orcid":"","institution":"Bishop's University","correspondingAuthor":true,"prefix":"","firstName":"Russell","middleName":"","lastName":"Butler","suffix":""}],"badges":[],"createdAt":"2025-07-12 12:38:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7108155/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7108155/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-16996-1","type":"published","date":"2025-09-26T15:57:56+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":87254257,"identity":"544e03b4-a86f-45bd-a35a-2106003b12a7","added_by":"auto","created_at":"2025-07-22 05:35:26","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3665645,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTemporal structure and inter-individual variability of behavioral and autonomic circadian rhythms. \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eA.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Unraveled time series from one subject (S36) showing coherent 24-h oscillations with ACC and IBI in anti-phase. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eB.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Grand mean and subgroup 24-h profiles reveal robust diurnal rhythms and sex-based differences in BPM but not ACC. Age comparisons showed non-significant reductions in rhythm amplitude among older adults. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eC.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003eACC acrophase consistently preceded BPM (mean lag: 1.78 h, p = 0.007). Males exhibited significantly later ACC acrophase than females, while no age-related phase differences were observed. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eD.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Individual 24-h profiles (n = 52) show consistent rhythmicity in most participants, with notable variability in amplitude and phase. Six participants were excluded due to poor fit or data sparsity.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"fig1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7108155/v1/255d1e529afaac9aecc17378.jpg"},{"id":87255128,"identity":"23c5be35-5543-42e3-a881-be37394700dd","added_by":"auto","created_at":"2025-07-22 05:43:26","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1396786,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDeterminants of behavioral–autonomic phase lag and circadian stability.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eExploration of inter-individual variability and predictors of the lag between ACC and BPM rhythms. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eA.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Distribution of ACC–BPM acrophase lags by sex shows no significant difference. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eB.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Hourly correlation between nighttime BPM and lag reveals a sinusoidal association. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eC.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Case examples highlight variation in inter-day ACC rhythm consistency, supported by heatmaps and correlation matrices. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eD–G.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Regression analyses show a significant negative correlation between nighttime BPM and lag (r = –0.55, p \u0026lt; 0.001); other candidate predictors (ACC consistency, BPM variability) were non-significant. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eF–I.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003eACC and BPM acrophases were not significantly correlated across subjects; lag was driven primarily by variability in ACC timing. ACC exhibited significantly greater day-to-day variability than BPM, indicating stronger stability in autonomic rhythms.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"fig2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7108155/v1/3c87ed108f0280ee1e4f38ee.jpg"},{"id":87254273,"identity":"7b064efb-38e4-4510-82ab-73027803329f","added_by":"auto","created_at":"2025-07-22 05:35:27","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":833896,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBehavioral lead and causal asymmetry in circadian rhythms.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eEvidence for behavior-first coupling between activity (ACC) and autonomic output (BPM).\u003cbr\u003e\n \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eA.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Pooled acrophase values show ACC peaks significantly earlier than BPM (Δ = 2.2 h, p \u0026lt; 0.0001). \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eB.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003eDay-to-day acrophase variability was similar across signals. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eC.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Higher nighttime BPM (1–4 AM) was associated with more negative ACC–BPM lag (p \u0026lt; 0.0001). \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eD.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003eWithin-subject analyses showed ACC acrophase better predicted daily lag than BPM, though marginally. \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eE.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Daytime ACC correlated more strongly with that-night BPM than the reverse, indicating temporal asymmetry (p = 0.0039). \u003c/em\u003e\u003cem\u003e\u003cstrong\u003eF.\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Granger causality analysis confirmed more frequent and significant causal flow from ACC to BPM across subjects (p = 0.0045), supporting a behavior-first model of circadian coordination.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"fig3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7108155/v1/0e85b9c9767327fafc3006eb.jpg"},{"id":92430576,"identity":"247d709f-a61f-402b-a88b-d947ca97e462","added_by":"auto","created_at":"2025-09-29 16:06:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6756215,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7108155/v1/0f985051-302d-4459-ac8e-46f529afa871.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Locomotor-Autonomic Coupling is Phase-Lagged and Behaviorally Driven: Evidence from long-term simultaneous heartrate and activity monitoring","fulltext":[{"header":"Introduction","content":"\u003cp\u003eCircadian rhythms govern a wide range of physiological and behavioral processes, including sleep\u0026ndash;wake cycles, core body temperature, hormone secretion, locomotor activity, and autonomic function. In humans, these rhythms are coordinated by a central pacemaker in the suprachiasmatic nucleus (SCN), yet they are expressed across multiple organ systems with varying degrees of independence and synchrony. Recent advances in wearable sensing technology have enabled large-scale, longitudinal monitoring of circadian patterns in free-living conditions, offering unprecedented resolution into human rhythms outside the laboratory.\u003c/p\u003e\n\u003cp\u003eSeveral recent studies have leveraged wearable data to quantify circadian physiology at population scale. For instance, Natarajan et al. [1] analyzed over 19,000 individuals using Fitbit data and identified consistent 24-hour rhythms in heart rate and activity, revealing systematic phase differences between these signals and strong modulation by age and sex. Similarly, Shim et al. [2] demonstrated that circadian parameters derived from accelerometry are predictive of aging and healthspan in a large cohort, positioning these digital rhythms as biomarkers of physiological resilience. In a related effort, Huang et al. [3] introduced a computationally efficient method to estimate circadian phase from wearable heart rate and temperature data, validating its accuracy across both lab-controlled and real-world datasets.\u003c/p\u003e\n\u003cp\u003eMorin et al. [4] have emphasized the importance of behavioral-physiological alignment, framing circadian misalignment as a key contributor to adverse outcomes in sleep, metabolism, and mental health. However, the degree to which autonomic rhythms\u0026mdash;particularly heart rate\u0026mdash;are temporally coupled with behavioral signals such as movement remains incompletely understood. Prior work by Karimi et al. [5] used smartwatch data to characterize the longitudinal dynamics of heart rate and activity, showing rich variation across individuals and time, yet without explicitly quantifying their phase relationships. Similarly, Hallgrimsson et al. [6] developed machine learning models to learn personalized cardiovascular responses from wearable sensor data, but focused on predictive performance rather than circadian structure.\u003c/p\u003e\n\u003cp\u003eEmerging evidence suggests that commercial wearables can detect and compare multi-system rhythms\u0026mdash;including activity, heart rate, and core body temperature\u0026mdash;in home environments [7]. Yet, these studies often lack the temporal resolution and duration required to assess causal directionality or inter-system coordination. Leota et al. [8], using large-scale actigraphy data, found that sleep timing significantly influences next-day activity levels, highlighting the importance of temporal sequencing in behavioral regulation. Building on this, Kim et al. [9] extended phase modeling techniques to both heart rate and body temperature across more than 50,000 participant-days, confirming circadian structure but leaving open questions about inter-system alignment. Meanwhile, sleep detection via heart rate alone has been validated in naturalistic settings [10], underscoring the feasibility of deriving autonomic states from wearable data.\u003c/p\u003e\n\u003cp\u003eTogether, these findings highlight the need to move beyond descriptive analyses and toward a mechanistic understanding of how autonomic and behavioral rhythms interact over time. Specifically, it remains unresolved whether the autonomic system merely reflects behavioral patterns or exhibits independent, potentially leading dynamics. Resolving this question requires high-frequency, long-duration, multi-modal recordings capable of detecting both phase relationships and causal flow between physiological systems under real-world conditions.\u003c/p\u003e\n\u003cp\u003eIn this study, we leverage 30 days of high-resolution, simultaneous accelerometry and heart rate recordings from 52 individuals to examine the temporal structure, alignment, and causal relationship between behavioral and autonomic circadian rhythms in naturalistic conditions. By quantifying acrophase timing, day-to-day stability, and lag between systems, we test whether the autonomic system lags behind behavioral output, whether this lag varies across individuals, and whether the direction of influence is consistently behavior-first. Our goal is to resolve whether the heart follows the body\u0026mdash;or leads it\u0026mdash;in daily life, and to uncover the individual and physiological factors that shape this internal coordination.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003eFifty-two healthy young adults (26 male, 26 female; age range: 18\u0026ndash;38 years) were recruited from the Bishop\u0026rsquo;s University and Universit\u0026eacute; de Sherbrooke communities. All participants provided written informed consent in accordance with ethics approval from the \u003cem\u003eCentre int\u0026eacute;gr\u0026eacute; universitaire de sant\u0026eacute; et de services sociaux de l\u0026rsquo;Estrie \u0026ndash; Centre hospitalier universitaire de Sherbrooke (CIUSSS de l\u0026rsquo;Estrie \u0026ndash; CHUS)\u003c/em\u003e. \u003cstrong\u003eAll methods were performed in accordance with the relevant guidelines and regulations.\u003c/strong\u003e Participants were screened to exclude any neurological or psychiatric disorders, sleep disorders, or use of medications affecting the central nervous system.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSmartwatch Data Acquisition\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEach participant wore a \u003cstrong\u003eSamsung Galaxy Watch Active2\u003c/strong\u003e continuously over a 30-day monitoring period. This device features:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003e8-channel green LED photoplethysmography (PPG) sensor array\u003c/li\u003e\n \u003cli\u003e3-axis accelerometer and gyroscope\u003c/li\u003e\n \u003cli\u003e360 x 360 Super AMOLED display (1.4\u0026quot;)\u003c/li\u003e\n \u003cli\u003eTizen OS (programmable environment with direct sensor access)\u003c/li\u003e\n \u003cli\u003eBattery capacity: 340 mAh (approximately 36 hours per charge)\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eCustom firmware written in C was deployed on the watches to enable continuous data sampling in energy-efficient bursts. Specifically, the accelerometer and PPG sensors were activated for \u003cstrong\u003e1 minute every 10 minutes\u003c/strong\u003e, recording at \u003cstrong\u003e10 Hz\u003c/strong\u003e, resulting in \u003cstrong\u003e4,320 minutes of data\u003c/strong\u003e per subject. This sparse sampling protocol allowed high-frequency physiological capture with minimal battery strain.\u003c/p\u003e\n\u003cp\u003eParticipants were instructed to wear the device at all times, including overnight, except during charging. Charging was typically done once daily during low-activity windows and took under 30 minutes. Subjects were encouraged to prioritize nighttime wear due to the reduced motion artifact and superior PPG signal quality during sleep. Adherence was verified visually during postprocessing.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cbr\u003e\u0026nbsp;Preprocessing and Hourly Matrix Construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRaw smartwatch sensor data were processed to extract hourly-resolved time series of accelerometry and interbeat interval (IBI) values for each subject. For each of the 52 participants, all available 10 Hz sensor data (accelerometer and photoplethysmography [PPG]) were loaded from local CSV files and segmented into non-overlapping 10-second epochs.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAccelerometry:\u003c/strong\u003e\u003cbr\u003eFor each epoch, the Euclidean norm of the 3-axis accelerometer signal was computed to yield a scalar magnitude of movement. These values were binned by local clock time into a 24 (hour) \u0026times; \u003cem\u003en\u003c/em\u003e (day) matrix, where each cell represents the mean accelerometer magnitude for that hour and day.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIBI Estimation:\u003c/strong\u003e\u003cbr\u003eThe raw PPG signal was bandpass-filtered (0.35\u0026ndash;4.0 Hz) and resampled to 30 Hz to improve peak detection fidelity. Epochs were accepted for IBI estimation only if they passed a signal quality check based on normalized autocorrelation, ensuring the presence of physiologically plausible periodicity. For accepted epochs, peaks were detected using amplitude thresholding and minimum peak distance criteria. Instantaneous IBI values (in seconds) were calculated as the mean time between adjacent peaks. These were also aggregated into a 24 \u0026times; \u003cem\u003en\u003c/em\u003e matrix per subject, where each cell represents the average IBI for a given hour and day. IBI were converted to BPM for further analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTime Alignment and Output:\u003cbr\u003e\u003c/strong\u003eTimestamps for all sensor data were converted to Eastern Time (EST/EDT) using timezone-aware conversion to ensure consistent alignment with local clock hour. Because both accelerometry and IBI were measured concurrently using the same device, no temporal alignment between behavioral and autonomic data streams (same timestamps applied to both signals). Only valid epochs with usable sensor data were included. Each subject\u0026rsquo;s data were stored as a dictionary containing a list of dates and two matrices (acc, ibi), each with shape (24 \u0026times; \u003cem\u003en_days\u003c/em\u003e). All subjects\u0026rsquo; matrices were saved into a single pickle file for downstream analysis\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRhythm Analysis, Modeling, and Statistical Inference\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCircadian Feature Extraction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFollowing preprocessing and alignment of the 24\u0026times;N (hour \u0026times; day) matrices for accelerometry (ACC) and inter-beat interval (IBI), we conducted subject-level and population-level analyses to quantify circadian structure, acrophase timing, and inter-system alignment.\u003c/p\u003e\n\u003cp\u003eFor each subject, mean hourly ACC and IBI values were computed across valid days, and IBI was converted to beats per minute (BPM) as BPM = 60 / IBI. To characterize 24-hour rhythms, we fit a fundamental cosinor model to each signal:\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ey(t) = \u0026beta;₀ + \u0026beta;₁ \u0026middot; cos(2\u0026pi;t / 24) + \u0026beta;₂ \u0026middot; sin(2\u0026pi;t / 24)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ewhere:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003e\u003cstrong\u003e\u0026beta;₀\u003c/strong\u003e is the mesor (mean level),\u003c/li\u003e\n \u003cli\u003e\u003cstrong\u003e\u0026beta;₁, \u0026beta;₂\u003c/strong\u003e are the cosine and sine amplitudes.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eusing ordinary least squares. Acrophase was derived from the fitted coefficients as:\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eϕ = (atan2(\u0026beta;₂, \u0026beta;₁) mod 2\u0026pi;) \u0026middot; (24 / 2\u0026pi;)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePhases occurring before 06:00 were wrapped forward by 24 hours to enable consistent comparison across subjects and systems. Per-subject cosinor fits were overlaid with raw 24-hour mean signals and visually inspected to identify poor fits or sparse coverage (N=6 subjects excluded from downstream analyses).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGroup-Level Comparisons\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo assess sex and age differences in circadian profiles, subjects were grouped by sex and age (\u0026lt;30 vs. \u0026ge;30 years), and group-wise grand mean \u0026plusmn; standard error curves were plotted. At each hour, unpaired \u003cem\u003et\u003c/em\u003e-tests were used to compare group means, with statistically significant hours annotated.\u003c/p\u003e\n\u003cp\u003eDifferences in acrophase timing across groups (sex, age) and modalities (ACC vs. BPM) were evaluated using either paired or unpaired \u003cem\u003et\u003c/em\u003e-tests, as appropriate. In all tests, normality assumptions were visually verified. Effect sizes (mean differences in hours) and test statistics (t, p) were reported.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCircadian Coupling Metrics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo examine coupling between locomotor (ACC) and autonomic (BPM) rhythms, we computed daily acrophases for each signal and calculated per-day lag as:\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026Delta;ϕ_lag = ((ϕ_ACC - ϕ_BPM + 12) mod 24) \u0026ndash; 12\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSubject-level lag distributions were compared across sex and correlated with physiological features, including nighttime BPM (1\u0026ndash;5 AM), 24-h BPM standard deviation, and locomotor rhythm consistency (mean pairwise inter-day Pearson \u003cem\u003er\u003c/em\u003e). ACC consistency was computed from the lower triangle of inter-day correlation matrices derived from the daily ACC profiles.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eVariability and Predictive Models\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo assess variability in acrophase timing across days, we computed the standard deviation of daily acrophases per subject for each signal. Paired \u003cem\u003et\u003c/em\u003e-tests were used to compare across signals. Additionally, for each subject, we fit linear models predicting daily lag from either ACC or BPM acrophase, computing R\u0026sup2; as the proportion of variance explained. Paired \u003cem\u003et\u003c/em\u003e-tests across subjects compared R\u0026sup2; values for the two predictors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTemporal Asymmetry and Causal Directionality\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe investigated temporal asymmetry in autonomic\u0026ndash;behavioral coupling via two approaches:\u003c/p\u003e\n\u003col start=\"1\" type=\"1\"\u003e\n \u003cli\u003e\u003cstrong\u003eDay-to-night correlation asymmetry\u003c/strong\u003e: For each subject, we correlated:\u003cul type=\"circle\"\u003e\n \u003cli\u003ePrevious-day daytime ACC (9\u0026ndash;17h) with same-night BPM (1\u0026ndash;4h)\u003c/li\u003e\n \u003cli\u003eNighttime BPM with next-day ACC\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eMean correlations were compared using a paired \u003cem\u003et\u003c/em\u003e-test.\u003c/p\u003e\n\u003col start=\"2\" type=\"1\"\u003e\n \u003cli\u003e\u003cstrong\u003eGranger causality\u003c/strong\u003e: We tested for unidirectional influence (lag=1) between daytime ACC and nighttime BPM using the grangercausalitytests function from statsmodels. Two models were tested per subject:\u003cul type=\"circle\"\u003e\n \u003cli\u003eACC[t-1] \u0026rarr; BPM[t]\u003c/li\u003e\n \u003cli\u003eBPM[t-1] \u0026rarr; ACC[t]\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eDistributions of p-values were compared using the Wilcoxon signed-rank test to assess group-level asymmetry.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eVisualization and Summary Statistics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data visualization was performed using matplotlib, seaborn, and numpy. Histogram bins, violin plots, scatter plots with regression fits, and subject-level heatmaps were used to illustrate rhythmic profiles, acrophase variability, inter-system lag, and consistency metrics.\u003c/p\u003e\n\u003cp\u003eAll statistical analyses were performed in Python using:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003escipy.stats for \u003cem\u003et\u003c/em\u003e-tests, Pearson correlations, standard deviations\u003c/li\u003e\n \u003cli\u003estatsmodels for Granger causality\u003c/li\u003e\n \u003cli\u003esklearn.linear_model.LinearRegression for acrophase-lag R\u0026sup2; computation\u003c/li\u003e\n \u003cli\u003ecircmean and circstd from scipy.stats for circular statistics\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eStatistical significance was assessed at \u0026alpha; = 0.05 unless otherwise specified, and \u003cem\u003ep\u003c/em\u003e-values were corrected where relevant (e.g., per-hour testing) through visual annotation but not formal multiple-comparisons correction.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the temporal structure, group-level dynamics, and individual variability in circadian rhythms derived from accelerometry (ACC) and heart rate (BPM) data in our full cohort (n\u0026thinsp;=\u0026thinsp;52). Panel A shows an unraveled, 44-day time series of z-scored ACC and IBI (inverse of BPM) in a single subject (S36) revealing coherent circadian oscillations with substantial inter-day modulation. ACC (blue) and IBI (orange) exhibit tight anti-correlated oscillations, with activity peaking during day (white background) and IBI peaking during night (gray background). Panel B1 displays grand mean 24-h rhythms, highlighting a robust diurnal pattern for both signals, peaking in the late afternoon. Sex differences are shown in B2\u0026ndash;B3, with females exhibiting significantly elevated early-morning BPM compared to males (* p\u0026thinsp;\u0026lt;\u0026thinsp;0.05), while ACC showed a modest trend but did not reach significance. Age differences (Panels B4\u0026ndash;B5) revealed lower activity and BPM in older adults (\u0026ge;\u0026thinsp;30 years) compared to younger adults (\u0026lt;\u0026thinsp;30), particularly in the early afternoon, but age differences were not significant in this cohort.\u003c/p\u003e\u003cp\u003eBar plots in Panels C1\u0026ndash;C5 quantify acrophase differences across signals and subgroups. ACC rhythms peaked significantly earlier than BPM rhythms, with a mean phase difference of 1.78 hours (C1; t\u0026thinsp;=\u0026thinsp;2.80, p\u0026thinsp;=\u0026thinsp;0.007), indicating a temporal lag between behavioral activation and subsequent autonomic arousal. Males exhibited a significantly later acrophase than females in ACC signal (C2; Δ\u0026thinsp;=\u0026thinsp;2.20h, t\u0026thinsp;=\u0026thinsp;2.12, p\u0026thinsp;=\u0026thinsp;0.040), but not BPM (C3; p\u0026thinsp;\u0026gt;\u0026thinsp;0.2). No significant age-related differences in acrophase were observed for either ACC (C4; p\u0026thinsp;=\u0026thinsp;0.948) or BPM (C5; p\u0026thinsp;=\u0026thinsp;0.966). Together, these findings suggest a modest but reliable lag between motor and cardiac rhythms, a sex difference in circadian timing, and stable phase alignment across age groups.\u003c/p\u003e\u003cp\u003ePanel D presents raw individual-level 24-hour profiles of accelerometry (ACC, blue) and heart rate (BPM, red) across all 52 participants. Solid lines depict the mean time series, while dashed lines show fitted cosinor curves with corresponding acrophase markers (dots). Most participants exhibited coherent circadian patterns in both signals, with moderate alignment in acrophase and waveform morphology. However, six participants (highlighted in red: s14, s31, s42, s8, s3, s18) demonstrated insufficient data coverage or unreliable cosinor fits and were therefore excluded from further acrophase and rhythm strength analyses. This visual overview underscores substantial inter-individual variability in rhythm amplitude, phase, and coherence across behavioral and autonomic domains.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e investigates the inter-individual variability and determinants of lag between behavioral (ACC) and autonomic (BPM) rhythms. Panel A shows the distribution of ACC\u0026ndash;BPM acrophase lag across male/female split, males and females showed overlapping lag distributions and females showing a slightly longer lag (non-significant, p\u0026thinsp;=\u0026thinsp;0.276). Panel B examines the hour-by-hour inter-subject pearson correlation between hourly BPM and lag, revealing a sinusoidal correlation pattern (subjects with greater physiological lag (more negative) had higher BPM at night).\u003c/p\u003e\u003cp\u003ePanel C highlights the heterogeneity in locomotor circadian stability using two exemplar participants\u0026mdash;one with the highest inter-day consistency and one with among the lowest. The left plots show daily ACC heatmaps by hour, and the right-side plots show the corresponding inter-day ACC correlation matrices. High-consistency subjects show clear diurnal structure and high inter-day correlation, whereas low-consistency individuals show noisy or inconsistent patterns.\u003c/p\u003e\u003cp\u003eRegression analyses in panels D\u0026ndash;G explore predictors of ACC\u0026ndash;BPM lag. A robust negative correlation was observed between nighttime BPM and lag (r = \u0026minus;\u0026thinsp;0.55, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), indicating that higher nocturnal BPM was associated to poorer autonomic/locomotor alignment. However, inter-day ACC consistency (E) and BPM variability (G) were not significantly associated with lag.\u003c/p\u003e\u003cp\u003ePanels F\u0026ndash;H assess the synchrony and variability of ACC and BPM acrophases. Individual ACC and BPM acrophases were not correlated across subjects (F; r\u0026thinsp;=\u0026thinsp;0.02, p\u0026thinsp;=\u0026thinsp;0.901), and the lag between autonomic and locomotor activity was due almost entirely to variability in locomotor activity (r\u0026thinsp;=\u0026thinsp;0.84, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), confirming that the lag reflects a shifted locomotor rhythm rather than shifted autonomic response. Supporting this, panel I demonstrates that ACC acrophase variability across days was significantly greater than BPM variability, reflecting stronger rhythm regularity in the autonomic system.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e provides converging evidence that behavioral activity precedes and potentially regulates autonomic rhythms, based on pooled and intra-subject analyses of circadian acrophase and causal inference. Panel A shows the overall distribution of acrophase values for ACC and BPM (via IBI), pooled across all days and subjects. ACC rhythms peak significantly earlier than BPM (mean\u0026thinsp;=\u0026thinsp;14.6h vs. 16.8h, t = \u0026minus;\u0026thinsp;14.90, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), confirming a consistent behavioral lead. Despite this phase offset, the inter-day variability of acrophase (Panel B) was comparable between signals (mean SD\u0026thinsp;=\u0026thinsp;1.96h for ACC vs. 1.69h for BPM, t\u0026thinsp;=\u0026thinsp;1.28, p\u0026thinsp;=\u0026thinsp;0.209), indicating that both systems exhibit relatively stable phase timing across days within subjects. When pooled across all days and subjects, higher nighttime BPM (1\u0026ndash;4 AM) was associated with a more negative ACC\u0026ndash;BPM lag (Panel C, r = \u0026minus;\u0026thinsp;0.22, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), consistent with prior findings that elevated nocturnal heart rate reflects a longer lag between activity and subsequent BPM changes.\u003c/p\u003e\u003cp\u003ePanel D compares within-subject predictive power (r\u0026sup2;) of daily acrophase values for explaining daily ACC\u0026ndash;BPM lag within each subject. ACC acrophase more strongly predicted the magnitude and direction of the daily lag than BPM acrophase, though the difference approached but did not reach significance (t\u0026thinsp;=\u0026thinsp;1.75, p\u0026thinsp;=\u0026thinsp;0.090). To probe temporal directionality, panels E and F evaluate causal asymmetry in the relationship between behavior and autonomic output. Panel E shows that daily activity levels more strongly correlate with that-night BPM (ACC\u0026rarr;BPM) than the reverse (nightly BPM correlating with next-day acc, BPM\u0026rarr;ACC), with a significant paired t-test (p\u0026thinsp;=\u0026thinsp;0.0039). Granger causality analysis (Panel F) supports this interpretation: a larger number of participants exhibited significant unidirectional Granger-causality from daytime ACC to that-night BPM (yellow bars at low p-values), with a Wilcoxon test confirming this asymmetry (p\u0026thinsp;=\u0026thinsp;0.0045). Together, these findings suggest a behavior-first model of internal coupling, where daytime activity shapes subsequent autonomic regulation rather than being driven by it.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur 30-day, high temporal resolution investigation of locomotor and autonomic rhythms provides compelling evidence that behavioral activity reliably precedes\u0026mdash;and likely shapes\u0026mdash;autonomic patterns, challenging the conventional notion that a central pacemaker independently governs both systems. We observed a robust phase lag (mean \u0026asymp; \u0026minus;\u0026thinsp;1.8 h) wherein activity acrophase precedes heart rate acrophase. This lead was consistent across individuals and largely unaffected by age or sex.\u003c/p\u003e\u003cp\u003eThis behavioral lead is aligned with individual-level findings showing that daily behaviors such as physical activity, diet, and stress can influence nocturnal heart rate variability and amplitude [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Moreover, our Granger causality (GC) analyses support this directional relationship, reinforcing earlier demonstrations of bidirectional, sleep-stage\u0026ndash;dependent brain\u0026ndash;heart coupling using EEG and ECG data [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. These GC results are further validated by prior methodological reviews that highlight the use of GC in modeling predictive flow between interconnected physiological systems, including cardiovascular and respiratory dynamics [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eWhile both activity and heart rate exhibited strong 24-hour rhythmicity, we found that the inter-individual stability of autonomic acrophase was significantly higher than that of locomotor rhythms. This asymmetry supports the hypothesis that autonomic outputs are more tightly constrained by internal timing mechanisms\u0026mdash;potentially including intrinsic cardiac oscillators\u0026mdash;than behavioral activity, which may be more influenced by social and environmental cues. This finding complements earlier work demonstrating that heart rate rhythms derived from wearable devices show greater consistency than other physiological indicators across days [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eFurther supporting the functional impact of desynchrony, we found that higher nighttime heart rate was associated with more negative lag values\u0026mdash;i.e., greater misalignment between activity and autonomic rhythms. This pattern replicates previous findings linking elevated nocturnal BPM with circadian disruption and autonomic hyperarousal [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], and supports its use as a potential digital biomarker for circadian misalignment in real-world populations. Large-scale studies have shown that circadian disruption, as inferred from heart rate patterns, is associated with elevated risk for mood disorders and poor mental health outcomes [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eOur study benefits from high-density, 30-day continuous wearable recordings, overcoming the temporal limitations of traditional laboratory studies. While many prior studies using wearables have focused on descriptive analyses or machine learning prediction, our integration of causal modeling places this work in the growing field of chronophysiological inference. This aligns with the broader aims of chronomedicine, where wearable-derived circadian metrics are now being explored for diagnostic, therapeutic, and monitoring applications [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eStill, several limitations deserve mention. First, while Granger causality captures directional predictive relationships, it does not establish mechanistic causation. Confounding variables\u0026mdash;such as light exposure, psychological state, or caffeine intake\u0026mdash;may jointly affect both behavior and autonomic outputs. Future research should integrate wearable data with contextual sensors and experimental manipulations (e.g., scheduled activity paradigms or forced desynchrony protocols) to test whether behavioral changes can shift or realign autonomic rhythms [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eSecond, our sample size (n\u0026thinsp;=\u0026thinsp;52), while rich in temporal resolution, is modest compared to population-scale datasets. Nonetheless, the within-subject nature of our analyses supports strong inferences about temporal coordination. Future studies in larger, more diverse cohorts will allow for exploration of demographic and clinical moderators of behavioral-autonomic coupling, including chronotype, shift work, and sleep disorders.\u003c/p\u003e\u003cp\u003eFinally, although our study focused on accelerometry and heart rate, a more complete understanding of circadian regulation will require integration of additional physiological signals\u0026mdash;including heart rate variability, core body temperature, hormonal profiles, and sleep staging\u0026mdash;many of which can now be noninvasively captured using modern wearables [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. A systems-level perspective, grounded in temporal modeling approaches [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], will be essential for mapping the multidimensional landscape of human circadian physiology.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eOur findings support a model in which daily behavioral activity drives and entrains autonomic rhythms. This opens the door for behaviorally timed interventions\u0026mdash;such as scheduled exercise or meal timing\u0026mdash;to realign internal rhythms, especially in populations vulnerable to circadian disruption. Furthermore, nighttime heart rate emerges as a simple, objective, and real-time biomarker of internal misalignment, detectable with inexpensive consumer-grade devices. Long-term wearable data in naturalistic conditions reveal a consistent pattern: the body leads, and the heart follows. These findings refine our understanding of inter-system circadian organization and provide an empirical basis for using behavior to manage internal synchrony.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u0026nbsp;\u003c/strong\u003ethis work was funded by the National Science and Engineering Research Council of Canada (NSERC) Discovery Grant program RGPIN1507.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u0026nbsp;\u003c/strong\u003eS.G. recruited participants and led data acquisition, including MRI and smartwatch recordings. O.D. developed the smartwatch firmware in C for raw data capture and extraction. MG, AZ, and CL performed statistical analyses and signal processing. R.B. conceived and supervised the study, provided funding, and oversaw all aspects of study design and interpretation. All authors contributed to manuscript preparation and approved the final version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests:\u0026nbsp;\u003c/strong\u003ethe authors declare no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability:\u0026nbsp;\u003c/strong\u003eAll raw and preprocessed data generated and analyzed during this study are available from the corresponding author ([email protected]) upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations and approval for human experiments:\u0026nbsp;\u003c/strong\u003eAll participants provided written informed consent in accordance with ethics approval from the \u003cem\u003eCentre int\u0026eacute;gr\u0026eacute; universitaire de sant\u0026eacute; et de services sociaux de l\u0026rsquo;Estrie \u0026ndash; Centre hospitalier universitaire de Sherbrooke (CIUSSS de l\u0026rsquo;Estrie \u0026ndash; CHUS)\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u0026nbsp;\u003c/strong\u003ethis work was funded by the National Science and Engineering Research Council of Canada (NSERC) Discovery Grant program RGPIN1507.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eNatarajan, A., Gleichauf, K., Khalid, M., Heneghan, C., \u0026amp; Schneider, L. D. (2025). Circadian rhythm of heart rate and activity: a cross-sectional study. \u003cem\u003eChronobiology Int.\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 108\u0026ndash;121.\u003c/li\u003e\n\u003cli\u003eShim, J., Fleisch, E., \u0026amp; Barata, F. (2024). Circadian rhythm analysis using wearable-based accelerometry as a digital biomarker of aging and healthspan. \u003cem\u003enpj Digit. Med.\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 146.\u003c/li\u003e\n\u003cli\u003eHuang, Q., Patel, N., \u0026amp; Forger, D. B. (2023). Efficient assessment of real-world dynamics of circadian rhythms in free-living conditions. \u003cem\u003eJ. R. Soc. Interface\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 20230030.\u003c/li\u003e\n\u003cli\u003eMorin, C. M., Carrier, J., Bastien, C., \u0026amp; Harvey, A. G. (2025). Circadian rhythms revealed: unraveling the genetic, physiological, and behavioral tapestry of the human biological clock and rhythms. \u003cem\u003eFrontiers in Sleep.\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 1544945.\u003c/li\u003e\n\u003cli\u003eKarimi, F., Amoozgar, Z., Reiazi, R., Hosseinzadeh, M., \u0026amp; Rawassizadeh, R. (2022). Longitudinal analysis of heart rate and physical activity collected from smartwatches. \u003cem\u003earXiv\u003c/em\u003e, abs/2211.08628.\u003c/li\u003e\n\u003cli\u003eHallgrimsson, H. T., Jankovic, F., Althoff, T., \u0026amp; Foschini, L. (2018). Learning individualized cardiovascular responses from large-scale wearable sensor data. \u003cem\u003earXiv\u003c/em\u003e, abs/1812.01696.\u003c/li\u003e\n\u003cli\u003eWu, F., Langer, P., Shim, J., Fleisch, E., \u0026amp; Barata, F. (2024). Comparative efficacy of commercial wearables for circadian rhythm home monitoring from activity, heart rate, and core body temperature. \u003cem\u003earXiv\u003c/em\u003e, abs/2404.03408.\u003c/li\u003e\n\u003cli\u003eLeota, J., Czeisler, C. A., Le, F., et al. (2025). Sleep duration and timing are associated with next-day physical activity: insights from two large-scale wearable sensor studies. \u003cem\u003eProc. Natl. Acad. Sci. U.\u003c/em\u003e\u003cem\u003e S.\u003c/em\u003e\u003cem\u003e A.\u003c/em\u003e \u003cstrong\u003e122\u003c/strong\u003e, e2420846122.\u003c/li\u003e\n\u003cli\u003eKim, D., Mayer, C., Lee, M. P., Choi, S. W., Tewari, M., \u0026amp; Forger, D. B. (2023). Efficient assessment of real-world dynamics of circadian rhythms in heart rate and body temperature from wearable data. \u003cem\u003eJ. R. Soc. Interface\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 20230030.\u003c/li\u003e\n\u003cli\u003ePerez-Pozuelo, I., et al. (2022). Detecting sleep outside the clinic using wearable heart rate devices. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 7956.\u003c/li\u003e\n\u003cli\u003eMatias, M., et al. (2022). What possibly affects nighttime heart rate? Conclusions from N-of-1 observational data. \u003cem\u003eDigit. Health\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eBashan, A., et al. (2022). Brain and brain\u0026ndash;heart Granger causality during wakefulness and sleep. \u003cem\u003eFront. Neurosci\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eTaguchi, S., et al. (2021). A method for characterizing daily physiology from widely used wearable devices, including HR circadian rhythms. \u003cem\u003ePatterns\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 100332.\u003c/li\u003e\n\u003cli\u003eArnaldi, D., et al. (2024). Wearables in chronomedicine and interpretation of circadian health. \u003cem\u003eDiagnostics (Basel)\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 327.\u003c/li\u003e\n\u003cli\u003ePichot, V., Corbier, C., \u0026amp; Chouchou, F. (2024). The contribution of Granger causality analysis to our understanding of cardiovascular homeostasis: from cardiovascular and respiratory interactions to central autonomic network control. \u003cem\u003eFront. Netw. Physiol.\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 1315316.\u003c/li\u003e\n\u003cli\u003eDurgin, J., \u0026amp; Heath, E. (2023). The cardiac circadian clock: implications for cardiovascular health. \u003cem\u003eJACC Basic Transl. Sci.\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 1\u0026ndash;13.\u003c/li\u003e\n\u003cli\u003eLee, M. P., Kim, D. W., Fang, Y., Kim, R., Bohnert, A. S. B., Sen, S., \u0026amp; Forger, D. B. (2024). The real-world association between digital markers of circadian disruption and mental health risks. \u003cem\u003enpj Digit. Med.\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 13.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7108155/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7108155/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe temporal coordination between behavioral and autonomic rhythms is a critical feature of circadian physiology, yet the precise alignment and causal structure of this coupling remain poorly characterized in free-living humans. Using long-term wearable data spanning 30 days from 52 individuals, we analyzed accelerometry (ACC) and heart rate (BPM, via IBI) to quantify circadian phase alignment, inter-day stability, and temporal directionality between locomotor and autonomic systems. Across individuals, behavioral activity rhythms consistently peaked earlier than autonomic rhythms (mean lag: –1.8h, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001), with the lag largely attributable to greater variability in locomotor phase. Despite this temporal dissociation, both signals exhibited coherent 24-hour patterns and relatively stable inter-day acrophases. Lag magnitude was negatively correlated with nighttime BPM (r = –0.55, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001), suggesting a link between autonomic hyperactivation and desynchrony. Crucially, behavioral acrophase more strongly predicted daily lag fluctuations than BPM acrophase, and causal analyses revealed asymmetric dependencies: same-day activity levels were significantly predictive of nighttime heart rate, whereas the reverse relationship was weaker and less consistent. Granger causality confirmed a predominant flow from ACC to BPM across subjects (\u003cem\u003ep\u003c/em\u003e = 0.0045). These findings establish that autonomic rhythms lag behind and are shaped by preceding behavioral activation, supporting a behavior-first model of internal circadian organization.\u003c/p\u003e","manuscriptTitle":"Locomotor-Autonomic Coupling is Phase-Lagged and Behaviorally Driven: Evidence from long-term simultaneous heartrate and activity monitoring","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-22 05:35:21","doi":"10.21203/rs.3.rs-7108155/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-05T06:40:05+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-03T16:19:43+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"246672656452170286016064827457127177508","date":"2025-07-29T05:54:58+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-27T08:08:24+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"187213493393210035054119050014935583110","date":"2025-07-17T12:03:51+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-17T11:25:45+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-17T11:20:58+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-07-17T08:29:47+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-15T14:27:22+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-07-15T13:40:59+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5009c6c5-d6f7-428b-876a-516fe5c7dabd","owner":[],"postedDate":"July 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":51882448,"name":"Biological sciences/Neuroscience"},{"id":51882449,"name":"Biological sciences/Physiology"},{"id":51882450,"name":"Biological sciences/Psychology"},{"id":51882451,"name":"Social science/Psychology"}],"tags":[],"updatedAt":"2025-09-29T16:01:35+00:00","versionOfRecord":{"articleIdentity":"rs-7108155","link":"https://doi.org/10.1038/s41598-025-16996-1","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-09-26 15:57:56","publishedOnDateReadable":"September 26th, 2025"},"versionCreatedAt":"2025-07-22 05:35:21","video":"","vorDoi":"10.1038/s41598-025-16996-1","vorDoiUrl":"https://doi.org/10.1038/s41598-025-16996-1","workflowStages":[]},"version":"v1","identity":"rs-7108155","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7108155","identity":"rs-7108155","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}