The clinical value of circadian biofeedback in chronic heart failure

Chronic heart failure (CHF) is a major health concern, affecting over 64 million people worldwide and contributing to significant morbidity, mortality, and healthcare costs.¹ Despite advances in pharmacological management, many patients experience persistent fatigue, impaired functional recovery, and reduced quality of life, often linked to autonomic imbalance and disrupted sleep–wake cycles.^2,3 Current treatment strategies mainly target hospitalization and mortality, but they frequently overlook restoration of physiological balance and overall wellbeing. Growing evidence highlights circadian disruption as both a cause and consequence of autonomic dysfunction in CHF.^4,5 Circadian rhythms regulate cardiovascular, metabolic, and neuroendocrine functions such as heart rate (HR), blood pressure, and autonomic tone. Disruptions may arise from external lifestyle-related factors like shift work, irregular sleep, travel, or nightly noise, as well as internal mechanisms in CHF, where rising filling pressures and reduced cardiac output activate sympathetic and Renin–angiotensin–aldosterone system pathways. This overstimulation impairs the suprachiasmatic nucleus, disrupting the internal clock and contributing to circadian misalignment, reduced nocturnal recovery, daytime fatigue, and increased risk of arrhythmia, rehospitalization, and mortality.^5–7 Interventions aimed at restoring circadian alignment may therefore represent a novel approach to strengthen cardiovascular resilience and promote behavioural stability in CHF.

Circadian regulation of cardiovascular physiology is mediated by a hierarchical clock system, in which the suprachiasmatic nucleus synchronizes peripheral clocks in the heart, vasculature, and autonomic nervous system through neural, humoral, and behavioural cues. This system governs 24 h rhythms in HR, blood pressure, myocardial contractility, vascular tone, and inflammatory signalling, shaping daily patterns of cardiovascular risk and recovery.^4,8,9 In CHF, these rhythms are typically dampened rather than abolished, suggesting preserved but weakened circadian control that may remain amenable to behavioural or therapeutic entrainment.^4,9,10

Multiple mechanisms contribute to circadian misalignment in CHF. Chronic sympathetic overactivity and reduced parasympathetic tone blunt normal day–night oscillations in autonomic function and heart rate variability, while altered visceral afferent feedback from the failing heart may disrupt central clock signalling and desynchronize peripheral cardiac clocks;^4,9 which is a strong predictor of adverse outcomes.^5,6,8,11 At the same time, sleep-related mechanisms play a critical role. Nocturnal fluid redistribution can exacerbate sleep-disordered breathing, including obstructive and central sleep apnoea, introducing intermittent hypoxia and recurrent arousals that provoke nocturnal sympathetic surges and further destabilize circadian regulation.^4,12

Beyond autonomic and sleep-related pathways, behavioural and environmental factors such as irregular timing of physical activity, daytime napping, meals, and light exposure may further desynchronize peripheral clocks from the central pacemaker, even in patients receiving optimal medical therapy.^4,13 Recent state-of-the-art reviews highlight that circadian disruption in cardiovascular disease arises from the interaction of molecular clock perturbations, autonomic imbalance, sleep disturbance, and lifestyle-related timing cues.^4,9,14 Together, these mechanisms provide a strong rationale for circadian-informed interventions that target daily behavioural timing as a complementary strategy to restore physiological rhythmicity in heart failure.

Recent advances in wearable sensing technologies enable continuous, real-world assessment of HR, HRV, physical activity, and sleep–wake patterns. When combined with analytical approaches such as cosinor modelling, these data allow characterization of individual circadian profiles and create opportunities to translate physiological rhythms into personalized, behaviourally actionable feedback.^6,15

Digital health tools that deliver personalized feedback have improved disease self-management across chronic conditions.^16,17 In CHF, however, current strategies focus mainly on symptom recognition and lifestyle guidance, such as weight monitoring, sodium restriction, and fluid control,¹⁶ while rarely considering the timing of behaviour or physiological rhythms. In CHF, where circadian rhythms are preserved but blunted,¹⁰ aligning activity, rest, and sleep with individual circadian profiles may enhance autonomic flexibility and recovery. Moderate physical activity and short, well-timed naps have both been shown to improve HRV and parasympathetic tone in healthy and athletic populations, whereas misaligned behaviours can destabilize rhythms and impair sleep quality.^18–23 Despite this promise, circadian-aligned interventions in CHF remain largely unexplored, with prior efforts focused mainly on pharmacologic chronotherapy or laboratory studies.^4,14 To date, no study has tested wearable-guided circadian feedback as a personalized self-management tool in this population. In this context, circadian biofeedback is defined as interpreted physiological feedback delivered at a behavioural timescale, rather than real-time closed-loop modulation of autonomic function. Unlike classical HRV biofeedback, which trains autonomic responses through moment-to-moment breathing or relaxation control, circadian biofeedback translates longitudinal physiological rhythms into guidance on when to perform daily behaviours such as activity, rest, and sleep. This approach emphasizes temporal alignment with endogenous circadian patterns rather than acute physiological regulation.

This study evaluated the physiological effects and clinical value of a wearable-guided circadian feedback tool for patients with CHF. The primary objective was to determine whether circadian-aligned guidance on activity, rest, and sleep improves autonomic regulation across 24 h, nocturnal, and activity-related domains. Secondary objectives were patient-centred and behavioural outcomes, assessed by a structured questionnaire (study experience, adherence to specific advice, and perceived effects) and by actigraphy (sleep, activity).

This single-centre, crossover study investigated the effects of a wearable-guided circadian feedback intervention in adults (≥18 years) with stable CHF (NYHA class II-III) attending the cardiology outpatient clinic of the Máxima Medical Centre (Veldhoven and Eindhoven, the Netherlands). Patients were eligible regardless of heart failure aetiology or left ventricular ejection fraction. Recruitment was coordinated through treating cardiologists, after which the coordinating investigator provided verbal and written study information. Exclusion criteria included permanent atrial fibrillation (due to uncertain data quality), inability to perform daily physical activities (e.g. orthopaedic or neurological limitations), and inability to wear sensors (e.g. dermatological conditions). All participants provided written informed consent. The study protocol was reviewed by the local medical ethics committee of Máxima Medical Centre, which determined that the study did not fall under the Dutch Medical Research Involving Human Subjects Act (WMO) and issued a formal waiver indicating that prospective trial registration was not required (reference number N25.006). All procedures were conducted in accordance with the principles of the Declaration of Helsinki.

The study followed a 3-week protocol, detailed below.

Week 0: Inclusion. Eligible patients were enrolled during outpatient visits and provided with the monitoring devices.

Week 1: Baseline mapping. Baseline measures of HRV, sleep, activity, and circadian parameters were obtained. Circadian phase markers were extracted to identify each participant’s acrophase (optimal timing for a 30-min walk) and local nadir (optimal timing for a 20-min nap). Data collection was fully non-invasive and did not interfere with clinical care.

Week 2: General advice. Participants received standardized education on three self-management strategies: (i) a daily 30-min walk at moderate intensity and (ii) a 20-min nap. Education was delivered verbally during a home visit and supported by illustrated flyers (Appendix IV). Walking intensity was guided by the Borg Scale of Perceived Exertion (scores 4–6, ‘green zone’) to prevent undertraining or overexertion. For naps, participants were instructed to rest for exactly 20 min to maximize restorative benefits while avoiding deeper sleep stages. (iii) Basic sleep hygiene advice was also provided, focusing on reducing evening light exposure and maintaining consistent bed- and wake-times.

Week 3: Circadian biofeedback. Personalized recommendations were generated from each participant’s circadian profile, calculated using the mean timing of circadian rhythm parameters (acrophase, nadir, and sleep–wake transitions) across all recorded days of Week 1. Advice included (i) performing the daily walk around the individual acrophase (peak activity period), (ii) napping near the local daytime nadir (lowest HR outside the sleep window), and (iii) aligning sleep hygiene with individual bedtime and wake-up markers.

Week 4: Sensor return and end of-study questionnaire.

At finalization of the study patients returned the CircAlign-HF monitoring system and completed the end-of-study questionnaire.

Baseline characteristics were summarized using descriptive statistics. Continuous variables are reported as median [interquartile range (IQR)] and categorical variables as absolute counts with corresponding percentages, as data were not normally distributed.

Physiological metrics collected at Weeks 1, 2, and 3 are presented as median (IQR). Data distributions were assessed with the Shapiro–Wilk test, which indicated non-normality across variables. Accordingly, changes over time were analysed using the nonparametric Friedman test for repeated measures.

Post hoc pairwise comparisons were performed only when the Friedman test reached statistical significance. Comparisons were conducted for the following deltas: Week 2—Week 1 (ΔWk2–Wk1), Week 3—Week 1 (ΔWk3–Wk1), and Week 3—Week 2 (ΔWk3–Wk2), using Dunn’s test with Šidák correction to adjust for multiple comparisons. Statistical significance was defined as P < 0.05 (Šidák-corrected).

All analyses were conducted in SPSS (version 29.0) and MATLAB (version 2024b).

A total of 21 patients with CHF were enrolled, of whom 19 (90.5%) were male. The median age was 68 years (IQR 61–76), and the median BMI was 26.9 kg/m² (IQR 24.2–29.7). Median left ventricular ejection fraction was 45% (IQR 30–50.5), distributed across heart failure with reduced ejection fraction (HFrEF, 38.1%), heart failure with mildly reduced ejection fraction (HFmrEF, 14.3%), heart failure with preserved ejection fraction (HFpEF, 14.3%), and heart failure with improved ejection fraction (HFimpEF, 33.3%). The predominant aetiologies were ischaemic cardiomyopathy (33.3%), tachycardiomyopathy (28.6%), and dilated cardiomyopathy (14.3%). Most participants were in NYHA class II (95.2%), with only one in class III.

All patients received guideline-directed medical therapy, including universal use of SGLT2 inhibitors (100%). Most were treated with β-blockers (85.7%) and mineralocorticoid receptor antagonists (85.7%), while 57.1% received sacubitril/valsartan. Common comorbidities were atrial fibrillation (71.4%) and hypertension (33.3%). Device therapy was limited to two patients (9.5%), both with an implantable cardioverter-defibrillator. Full patient characteristics are presented in Table 2.

Adherence to the general behavioural recommendations was high. Most participants integrated the daily 30 min walk (68%) and the 20 min power nap (74%) into their daily routines during the intervention period. Adherence to circadian timed recommendations was more variable; approximately 47% of participants reported adjusting the timing of walks and naps to the suggested circadian windows. Common barriers included fixed work or family schedules, weather conditions, and pre-existing routines. These adherence patterns are further detailed in the qualitative analysis (Section 3.4).

This study evaluated the physiological and clinical value of the CircAlign-HF system, a wearable-guided circadian biofeedback tool for patients with CHF. We compared usual care to two levels of intervention: (i) general advice on daily rest, activity, and sleep hygiene and (ii) circadian biofeedback-guided recommendations aligning activity, rest, and sleep timing with individual circadian rhythms. General advice alone was sufficient to improve autonomic regulation, as reflected in increased vagal tone in 24 h HRV and transient HRV gains during exertion windows. The addition of chronotherapy principles produced a more targeted benefit, specifically enhancing nocturnal autonomic regulation through a progressive increase in parasympathetic activity at night. More than half of the participants also perceived improvements in sleep quality and daily energy distribution, reinforcing the subjective relevance of the intervention. Participants generally described the intervention as positive and easy to follow, with most integrating the daily walk and nap into their routines. Adherence to circadian-aligned recommendations was more variable, often constrained by external factors or pre-existing rhythms. Together, these findings highlight the potential clinical relevance of the CircAlign-HF system as a wearable-guided behavioural tool, combining short-term physiological modulation of autonomic regulation with perceived improvements in sleep quality and daily energy distribution.

The observed improvements in nocturnal HRV are most plausibly explained by enhanced parasympathetic activation and reduced sympathetic drive during sleep. Structured daytime activity combined with short, well-timed naps may have reduced evening sympathetic spillover and facilitated vagal predominance at night. This interpretation is consistent with experimental work in healthy adults, where naps timed to the circadian nadir acutely increase parasympathetic activity and cognitive performance, whereas misaligned, long naps impair sleep continuity.^21–23 In CHF, where sympathetic overactivity and blunted vagal tone are central features, even modest nocturnal restoration of vagal activity may aid myocardial recovery, improve diastolic filling, and reduce arrhythmic vulnerability by attenuating oxidative stress.^3,27 Anchoring bedtime to the HR downslope and wake time to the HR upslope provides a physiological scaffold for sleep–wake cycles, consistent with chronobiological principles^13,28 In CHF, better sleep quality and preserved nocturnal vagal activity are associated with improved metabolic function and prognosis,^3,12 highlighting the potential relevance of circadian-aligned behavioural strategies for long-term outcomes.

The transient improvements in diurnal activity-related HRV suggest that the behavioural advice itself, daily moderate-intensity walking and structured rest, was sufficient to improve autonomic flexibility, independent of circadian alignment. Exercise is known to accelerate vagal reactivation post-exercise and enhance baroreflex sensitivity, both key contributors to HRV recovery.^18,19 The absence of sustained effects in Week 3 may indicate that circadian alignment requires longer exposure or higher adherence to yield additive benefits. Partial adherence, reported by participants facing family or work constraints, likely also contributed to the attenuation and underscores the need to interpret these findings cautiously. Thus, both biological (autonomic modulation) and behavioural (structured routines) mechanisms appear to underlie the observed improvements.

Notably, objective actigraphy derived sleep metrics remained stable despite improvements in nocturnal HRV and perceived sleep quality. This apparent dissociation is not unexpected. Actigraphy primarily captures sleep duration and movement-based continuity and is limited in its ability to assess sleep depth or autonomic state.²⁵ In contrast, nocturnal HRV, particularly when assessed around the circadian nadir, reflects autonomic recovery and parasympathetic predominance associated with deeper, more restorative sleep.²⁹ It is therefore plausible that parasympathetic restoration and subjective sleep quality improved without detectable changes in total sleep time or sleep efficiency. Similar dissociations between subjective sleep quality, autonomic markers, and actigraphy-derived metrics have been reported in both sleep and cardiovascular research, including in heart failure populations,^30,31 underscoring the complementary rather than interchangeable nature of these measures.

Our findings contribute to the growing field of digital self-management and biofeedback in chronic disease. In CHF, most digital interventions have focused on telemonitoring of symptoms, weight, or blood pressure.³² Platforms such as TIM-HF2 have reduced hospitalizations through reactive, alert-based systems,³³ but these do not actively support restoration of physiological balance. Similarly, conventional cardiac rehabilitation emphasizes exercise and counselling,³⁴ which can improve HRV and functional capacity but rarely incorporate chronotherapy principles such as aligning daily behaviours with circadian rhythms.

More recently, proactive physiology-based tools have emerged. HRV biofeedback (HRVB) delivered via wearables can improve vagal tone and cardiovascular recovery by guiding resonance-frequency breathing³⁵ although remote HRVB trials have reported limited compliance.³⁶ These studies highlight both the promise and the practical barriers to scaling biofeedback interventions. In parallel, chronomedicine platforms such as TimeTeller integrate molecular and circadian data to individualize treatment timing,³⁷ but their reliance on biological sampling limits routine home implementation and these approaches have not yet been evaluated in CHF populations. Together, these developments illustrate both the potential of physiology-guided interventions and the practical challenges of translating them into scalable, home-based care.

When contextualized against this literature, the magnitude of HRV changes observed in the present study falls within the range reported for established non-pharmacological strategies. Exercise-based interventions in CHF typically report SDNN increases of approximately 15 to 40 ms and rMSSD increases of 10 to 30 ms, with considerable variability depending on baseline autonomic impairment, intervention duration, and the analytical window used.^38,39 HRV biofeedback interventions similarly demonstrate modest improvements in rMSSD and HF power, particularly in individuals with reduced baseline vagal tone, but these effects are often constrained by adherence in remote settings.⁴⁰

A key methodological consideration is that our analyses combined conventional 24 h HRV with short, circadian-timed windows of approximately 15 min at the activity- and HR–defined acrophase and nadir. These physiologically anchored windows are known to yield higher absolute HRV values and larger apparent week-to-week changes than full-day averages and should therefore not be interpreted as directly equivalent to 24 h Holter-derived metrics. Rather, they provide enhanced sensitivity to state-dependent autonomic modulation, particularly during sleep, which is increasingly recognized as a critical period for cardiovascular recovery and autonomic restoration.³

Against this backdrop, the CircAlign-HF system is, to our knowledge, the first to integrate continuous wearable monitoring with longitudinal circadian-aligned feedback on rest, activity and sleep in the home environment for patients with CHF. Unlike telemonitoring platforms that focus on risk detection or HRVB tools that train autonomic responses via breathing exercises, CircAlign-HF actively guides patients in structuring daily routines around their biological rhythms. Our findings show that such guidance can enhance nocturnal parasympathetic activity, counteracting the sympathetic overdrive central to CHF pathophysiology, and may promote more consolidated sleep and more stable daily activity rhythms.

Our findings open several avenues for future research.

First, clinical validation; larger trials with extended interventions and longer follow-up are required to establish whether circadian-aligned feedback produces sustained autonomic benefits and translates into meaningful outcomes such as fewer hospitalizations or improved functional capacity.

Second, refinement of the intervention; Our qualitative results suggest that strict alignment is not realistic for all patients. Incorporating flexibility in recommended time windows or embedding the algorithm into adaptive digital platforms may increase feasibility and adherence. Integration with existing digital health platforms could further enable real-time feedback, gamification, and remote monitoring, thereby enhancing engagement.

Third, targeted populations; Circadian biofeedback should also be evaluated in patients with more advanced CHF or comorbid circadian disruption (e.g. sleep apnoea, atrial fibrillation, or recent hospitalization for acute decompensation), where misalignment may be more pronounced, given that the present study primarily included NYHA class II patients receiving optimal medical therapy.

Fourth, mechanistic validation; Studies combining continuous wearable monitoring with molecular circadian markers (e.g. melatonin, cortisol) are needed to disentangle behavioural influences from central clock mechanisms. Extending the algorithm to detect early circadian destabilization could also enable proactive interventions and potentially prevent hospitalization, as demonstrated in prior telemonitoring trials.⁶

Fifth, data processing and scalability; Widespread implementation of circadian biofeedback will require scalable and transparent data-processing pipelines. Automated preprocessing of wearable data (e.g. artefact detection, wear-time validation, and missing-data handling), combined with algorithmic extraction of circadian timing markers such as acrophase, nadir, and rhythm amplitude, could enable efficient and reproducible personalization of feedback in larger populations. Importantly, these computational tools are intended to support clinicians and patients by translating physiological rhythms into interpretable timing recommendations, rather than functioning as autonomous decision systems.

Finally, broader applications; The concept of wearable-guided circadian biofeedback should be seen as a platform approach with relevance beyond CHF. By targeting autonomic imbalance and promoting structured daily rhythms, similar algorithms could be adapted to other chronic conditions (e.g. hypertension, diabetes, post-myocardial infarction recovery) and to psychiatric disorders such as depression, where daily routines are often disrupted.⁴¹ In this way, circadian biofeedback may evolve into a versatile digital health strategy for stabilizing physiology and behaviour across diverse diseases.

Key strengths of this study include its novel, patient-centred design, use of validated wearable sensors, and integration of circadian analysis with clinically interpretable feedback. The crossover design, with participants serving as their own control, enhanced sensitivity to within-subject changes. In addition, the mixed-methods approach, combining objective wearable-derived metrics with qualitative feedback, provided both mechanistic and patient-centred perspectives on feasibility and acceptability.

Limitations include the small sample size and short study duration, which constrained the ability to detect changes in behavioural outcomes or clinical endpoints. Moreover, the fixed-sequence, non-randomized design limits causal separation of time effects, general behavioural advice, and circadian-aligned recommendations, and precludes definitive attribution of observed changes to circadian alignment alone. The predominantly male, NYHA class II population may not be representative of women or patients with more advanced CHF. Adherence to circadian-aligned advice was variable, with many participants citing barriers to strict timing. Finally, the absence of biochemical or central circadian markers limited mechanistic validation of the observed physiological changes. Accordingly, findings should be viewed as hypothesis generating and supportive of physiological feasibility rather than clinical efficacy.

The datasets generated and analysed during the current study are not publicly available due to ethical and privacy restrictions, as participants did not provide consent for open data sharing. Pseudonymized data may be made available upon reasonable request to the corresponding author.