From molecule to bedside and back: the integration of computational and experimental physiology in the digital age

Historically, physiology has evolved from observational descriptions to the rigorous dissection of mechanisms. Today, we witness another transformation where experimental and computational methodologies are no longer separate disciplines, but rather interconnected strategies that are essential for navigating the nonlinear complexity of the heart. This Special Issue represents a maturation of the ‘Physiome Project’ vision (Loewe et al.; 2025): a quantitative, multi-scale description of physiological dynamics, now spanning from the metabolic state of a crossbridge to the haemodynamic flow in a fibrillating atrium. The contributions in this collection largely stem from the Cardiac Physiome Workshop 2024 in Freiburg, Germany. There, the community gathered to discuss the bidirectional crosstalk between theoretical foundations and specific clinical challenges, such as atrial fibrillation and drug safety. Indeed, we see a broadening of the use of models, which become predictive engines capable of exploring the design of novel therapies, the assessment of clinical risks in real time, and the scaling of physiology from molecule to organ. Thus, successful physiological insights now emerge from the iterative feedback between laboratory data and mathematical reconstruction. The interdisciplinary methodological frameworks highlighting this integration have been reviewed by Yang & Sobie (2025), who focus on defining the essential ‘ingredients’ for the effective integration of computational and experimental approaches. A central, recurring theme in this issue is the emergence of the digital twin. While this term is often used to describe patient-specific modelling in a clinical setting, several contributions in this collection emphasise that the concept also encompasses fundamental biophysical models. These mechanistic twins, though perhaps not tailored to a single individual, provide the necessary framework to interpret experimental variability and define the evolutionary robustness of cardiac function. This integration is crucially bidirectional: while computational models predict outcomes, high-resolution experimental and clinical mapping remains the basis that informs and constrains the next generation of in silico studies. In a seminal contribution, Noble (2026a) traces the theoretical foundations of our current understanding of cardiac rhythm and the models we use to simulate it. By reflecting on the evolution of pacemaker models, Noble demonstrates how mathematical frameworks did not just follow experimental discovery but often pointed toward the necessary experimental questions. This work is further contextualised by Maltsev et al. (2026), whose perspective highlights the evolutionary robustness of these physiological systems, which is a concept essential for understanding why certain rhythms persist or fail under pathological stress. The ongoing dialogue in this field is further reinforced by the commentary of Noble (2026b), which reminds us that, as our models become more complex, we must remain grounded in the fundamental biophysical principles that govern cardiac automaticity. Atrial fibrillation (AF) remains the most common sustained arrhythmia and its structural basis and maintenance mechanisms remain subjects of intense investigation. A major challenge in treating AF is the scale mismatch between the microscopic environment, where triggers arise, and the macroscopic mapping performed in clinical electrophysiology laboratories. Several works in this Special Issue explore how new high-resolution imaging and computational reconstruction technologies are bridging this divide, moving the field towards a more nuanced understanding of how structure influences function. Kulathilaka et al. (2026) exemplify the importance of atrial microstructure by implementing digital models derived from micron-resolution imaging of the human heart. They demonstrate that re-entrant drivers are not randomly distributed but are anchored to specific microstructural features. In an accompanying Perspective, Potse (2026) discusses how these findings suggest that AF drivers ‘like to live in narrow spots’, effectively linking the multiple wavelet and rotor hypotheses by showing how chaotic activity can be organised by anatomical constraints. This microstructural analysis is further explored in a Journal Club article by Cutler (2026), who examines the transition from microstructure to arrhythmia and the specific structural drivers that sustain re-entrant activity. Complementing these ex vivo insights, Horrach et al. (2026) provide a translational review of mapping techniques for evaluating fibrosis in patients. They discuss how various imaging modalities, beyond standard electroanatomic mapping, can characterise the pathological substrate in vivo, thereby bridging the gap between high-resolution histology and clinical catheter ablation. Beyond fixed structural obstacles, the dynamic, functional remodelling of ion channels provides a secondary, evolving substrate for arrhythmia. Meier et al. (2026) applied computational modelling to explore the rate-dependent trafficking of small-conductance calcium-activated potassium (SK) channels. The collection also points out that AF is not solely an electrical phenomenon but one with profound haemodynamic consequences. Bäck et al. (2026) demonstrate how virtual interventions, such as left atrial appendage occlusion, can predict improvements in blood flow and reduce stasis risk, offering a tool for pre-procedural planning. This work is discussed by Hassan et al. (2026). Perhaps the most direct clinical application of these ‘digital heart’ tools is seen in the prospective in silico trial by Dasí et al. (2026). By simulating drug effects across a population of models representing inter-individual variability, they identified combined SK and two-pore-domain potassium channel blockade as a potentially effective strategy for AF cardioversion. This approach illustrates how computational physiology can ‘de-risk’ pharmaceutical development by identifying synergistic targets and predicting outcomes before a single patient is enrolled in a trial. This approach further supports the translational promise of targeting the SK channel, as comprehensively reviewed by Zheng et al. (2026). In the ventricles, the focus shifts to the subtle interplay between activation sequences, repolarisation heterogeneity, and arrhythmia vulnerability. Refining the interpretation of non-invasive data, Smits et al. (2026) present evidence contrary to the assumption that late-activated regions always display shortened action potential durations to maintain T-wave concordance. To correctly interpret electrocardiographic imaging maps in patients with heart failure, they propose a computational strategy challenging longstanding clinical assumptions. The connection between physiological gradients and arrhythmia risk is further explored by Sobota et al. (2026), who show that the apicobasal repolarisation gradient diminishes or inverts with age. This loss of the natural gradient may paradoxically increase vulnerability to re-entrant arrhythmias, providing a potential mechanistic link between the ageing process and the broader societal challenge of managing arrhythmic risk in an increasingly ageing population. In safety pharmacology, models are proving increasingly valuable for predicting cardiotoxicity and minimising the risks associated with specific therapies. Villar-Valero et al. (2026) present a parametric study using three-dimensional (3D) magnetic resonance imaging-based models to predict the effects of doxorubicin. By computationally modelling the specific patterns of diffuse interstitial fibrosis induced by the drug, they identify this structural remodelling as a key biomarker that promotes conduction block. As noted by Durço et al. (2026) in the accompanying Journal Club, this illustrates how the passive collagen network dictates the propagation of electrical waves. Chabiniok & Zaha (2026) expand upon this theme by reviewing the application of cardiac digital twins in assessing chemotherapy-induced substrates. Mora et al. (2026) apply electromechanical simulations to assess the effectiveness of omecamtiv mecarbil in treating heart failure. As discussed by Malicki et al. (2025), this approach offers a well-controlled in silico environment to dissect drug effects in the absence of other systemic influences. The link between metabolism and mechanics is a critical determinant of dysfunction in conditions such as diabetic cardiomyopathy. Musgrave et al. (2026) address this phenomenon with metabolite-sensitive crossbridge models to investigate how the metabolic environment in diabetes directly alters myofilament energetics and stiffness. Such models are important in providing a rigorous thermodynamic framework to explain how changes in metabolite concentrations, often seen in diabetes, impact muscle mechanoenergetics at the molecular level. Complementing this subcellular perspective, Bertrand et al. (2026) review cardiac remodelling in type 2 diabetes across multiple scales, discussing how computational multiscale modelling can link these subcellular defects with tissue-level fibrosis and autonomic neuropathy, thereby improving our understanding of disease progression. At the cellular level, Folkmanaite et al. (2026) demonstrate the value of computational approaches by utilising electromechanical modelling of human-induced pluripotent stem cell-derived cardiomyocytes to predict drug-induced contractility effects. In structural heart disease, Dvoulety & Sitina (2026) propose a haemodynamic–energetic mechanism for sudden cardiac death in aortic stenosis. Their model describes a state of ‘cardiac stalling’, whereby the energy demand of the hypertrophied ventricle exceeds the supply during stress, leading to mechanical failure. Townsend (2026) emphasises that this energy-balance framework explains syncope that purely hydraulic models may miss, pointing out the ability of models to assess catastrophic events in silico that cannot be ethically induced in human subjects. Moreover, models are increasingly capable of predicting long-term structural adaptation. Sharifi et al. (2026) incorporated multiscale finite element modelling to elucidate the role of the baroreflex feedback loop in ventricular growth, connecting neural control to tissue mechanics. Moscoloni et al. (2026) further recognise the significance of considering biological variables frequently disregarded in generic models: sex alone explains at least 25% of ventricular morphological variability. These fundamental shape disparities between male and female hearts should be incorporated into future patient-specific or sex-specific simulations. Finally, the developmental origins of these conditions are explored by Bueno-Palomeque et al. (2026), who use a synthesis of clinical and computational data to characterise preadolescents born with intrauterine growth restriction. Moving from pharmacological triggers to structural substrates, this work demonstrates how multi-scale modelling can isolate the specific contributions of ionic remodelling to the developing heart, reinforcing the ‘developmental origins of health and disease’ hypothesis. The boundary at the interface between biological and digital systems is becoming increasingly blurred, creating fertile ground for the integration of bioelectronics and artificial intelligence. Stark et al. (2026) review the development of flexible, implantable systems that utilise real-time data processing to close the loop between sensing and therapeutic intervention. These autonomous, closed-loop bioelectronic platforms are capable of sensing and modulating cardiac function in real time and represent an advancement in translational computational physiology. Technological progress in this area is supported by the development of sophisticated cellular-level models, such as the voltage-modulated Markov chain model for the Piezo1 ion channel presented by Ogiermann et al. (2026). By refining our ability to simulate mechanotransduction, such models provide the biophysical foundation for the next generation of intelligent, mechanosensitive devices. Furthermore, Greiner et al. (2025) introduce a deep learning-enabled toolkit for the automated 3D segmentation of ventricular cardiomyocytes to build these models at scale, effectively removing the manual bottleneck in high-resolution structural reconstruction. Physiological principles are tested and validated further at the extremes of environmental stress. Tripoli et al. (2026) used multiscale models to assess cardiac function across the spectrum from microgravity to hypergravity, informing spaceflight medicine and our fundamental understanding of gravitational adaptation. As Collia et al. (2026) note in an accompanying Perspective, these simulations allow us to explore physiological states that cannot be replicated in standard laboratory settings, offering a unique perspective on the robustness of cardiac function. Finally, comparative physiology continues to provide essential evolutionary benchmarks for identifying conserved mechanisms of cardiac automaticity and regeneration. This collection includes a novel mathematical model of the zebrafish ventricular action potential by Cestariolo et al. (2026), delivering a quantitative framework for this increasingly popular model of cardiac development. In the sinoatrial node, Linder et al. (2026) examine how sympathetic stimulation can compensate for hypocalcaemia-induced bradycardia. By comparing human and rabbit cells, they identify fundamental mechanisms of pacemaking robustness that are conserved across species as well as noable differences, thereby further grounding our computational reconstructions in comparative biological reality. The reliability of any computational model is inherently linked to the quality and resolution of the experimental data used for its construction and validation. As the complexity of simulations increases, so too does the requirement for highly accurate ground-truth datasets for parametrisation and validation. This collection showcases significant advancements in integrated experimental setups that allow for the comprehensive analysis of cardiac function across scales. Siles et al. (2026) describe a sophisticated platform for integrated panoramic optical and electrical mapping, enabling detailed arrhythmia studies in isolated hearts with high spatial and temporal resolution. This approach provides the necessary validation data for complex 3D ventricular models, a topic expanded upon by Zgierski-Johnston & Huisken (2025). They discuss the profound implications of being able to ‘see’ the heart in 3D in order to analyse the mechanical and electrical drivers of arrhythmia. However, even the most advanced experimental data are subject to variability and measurement error. Acknowledging that data is never ‘perfect’, Tanner et al. (2026) present a rigorous framework for uncertainty quantification in cardiac modelling. Focusing specifically on myocardial fibre orientation and activation patterns, they argue that incorporating error bars into simulation predictions is not simply a statistical exercise but a fundamental requirement for establishing the reliability and clinical safety of in silico results. By treating parameter uncertainty with the same rigour as experimental error, the research community can move towards more robust, reproducible, and ultimately translatable predictive models. The papers collected in this Special Issue demonstrate that computational physiology has evolved from a niche subdiscipline into a central pillar of modern cardiovascular research. The ability to integrate experimental data and computational simulations, spanning the scale from the molecular energetics of a crossbridge to the complex haemodynamic flows of a fibrillating atrium, is now a reality, not a distant aspiration. This convergence is accelerating the translation of fundamental physiological principles into clinical understanding and, ultimately, novel therapeutic strategies. As we look to the future, several key themes have emerged as priorities for the community. The first of these is the continued refinement and precise application of digital twin technology. While patient-specific modelling shows great potential for precision medicine, the contributions in this issue remind us that digital twins of in vitro experiments and generic mechanistic models, which explain fundamental biological principles, remain indispensable for interpreting experimental variability and defining the evolutionary robustness of cardiac function. Secondly, high-resolution, multi-modal experimental optical, ultrasound, and structural mapping must remain at the core of our field. This collection highlights that the next generation of in silico tools will only be as robust as the experimental data used to constrain them. Additionally, as our models move closer to the bedside, a rigorous commitment to uncertainty quantification and the inclusion of biological diversity, including age, sex, and developmental origins, is no longer optional; it is a key requirement for clinical safety and efficacy. Furthermore, the integration of machine learning and bioelectronics is enhancing our ability to reconstruct cardiac structure and modulate function in real time. The adoption of automated image analysis and autonomous closed-loop systems represents a practical shift towards more responsive physiological monitoring that remains grounded in the biophysical reality of the heart as a complex multicellular network. We hope that this Special Issue serves as both a comprehensive resource for the current community and a catalyst for the next generation of researchers. By continuing to close the gap between the analogue world and its digital representation, we can ensure that cardiac physiology remains at the forefront of the quest to understand, diagnose, treat, and ultimately prevent cardiovascular disease. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. No competing interests declared. V.T.: Conception or design of the work; Drafting the work or revising it critically for important intellectual content; Final approval of the version to be published; Agreement to be accountable for all aspects of the work. A.L.: Conception or design of the work; Drafting the work or revising it critically for important intellectual content; Final approval of the version to be published; Agreement to be accountable for all aspects of the work. The Cardiac Physiome Workshop 2024 in Freiburg, Germany, was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project number 539 837 774. VT was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project number 403 222 702 – SFB 1381, the Cluster of Excellence ‘Centre for Integrative Biological Signalling Studies’ (CIBSS) by the DFG under Germany's Excellence Strategy – EXC-2189 – 390 939 984, by the Hans A. Krebs Medical Scientist Programme, Faculty of Medicine, University of Freiburg, and by the Baden-Württemberg Stiftung—ArrhythMEC. AL acknowledges support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project number 507 828 355 and by the Leibniz ScienceCampus ‘Digital Transformation of Research’ with funds from the programme ‘Strategic Networking in the Leibniz Association’.