Pulmonary hypertension (PH) is a progressive and often fatal disease characterized by elevated blood pressure in the pulmonary arteries. Despite advances in pharmacotherapy and interventional techniques, diagnosis remains challenging—patients are frequently diagnosed late, when irreversible vascular remodeling has already occurred. Accurate, early diagnosis and individualized treatment planning are critical to improving outcomes. Over the past decade, computational simulation of pulmonary circulation has emerged as a powerful tool that can model blood flow dynamics with unprecedented detail. By integrating patient-specific imaging data and hemodynamic measurements, these simulations offer clinicians a virtual window into the pulmonary vasculature, enabling them to test hypotheses, predict disease progression, and optimize interventions before ever touching a patient. This article explores the current state and future potential of computational simulation in the diagnosis and treatment of pulmonary hypertension, emphasizing how it is reshaping cardiovascular medicine.

The Physiology of Pulmonary Circulation

The pulmonary circulation is a low-pressure, high-compliance system that efficiently delivers deoxygenated blood to the alveolar capillaries for gas exchange and returns oxygenated blood to the left atrium. Understanding its normal mechanics is essential to appreciating how PH develops and how simulation can capture pathological changes.

Normal Pulmonary Hemodynamics

In a healthy adult, mean pulmonary artery pressure (mPAP) ranges from 12 to 16 mmHg, with pulmonary vascular resistance (PVR) typically below 3 Wood units. The pulmonary arteries are highly distensible, allowing them to accommodate the entire cardiac output with minimal pressure elevation. Wave propagation and reflection occur along the arterial tree, influenced by vessel stiffness and branching geometry. This delicate balance is maintained by endothelial function, smooth muscle tone, and extracellular matrix composition.

Pathophysiology of Pulmonary Hypertension

Pulmonary hypertension arises from a complex interplay of vasoconstriction, endothelial dysfunction, thrombosis in situ, and progressive vessel remodeling. The resultant increase in PVR and mPAP (>25 mmHg at rest) imposes a chronic afterload on the right ventricle, leading to hypertrophy, dilation, and eventually right heart failure. Classification distinguishes five groups, with pulmonary arterial hypertension (PAH, Group 1) and chronic thromboembolic disease (Group 4) being particularly amenable to computational modeling due to their focal or diffuse vascular abnormalities. Simulation can quantify how these structural changes alter flow patterns, pressure gradients, and shear stress distributions along the endothelium.

Computational Simulation Approaches

Modern computational models of pulmonary circulation range from detailed three-dimensional fluid dynamics simulations to simplified lumped-parameter networks. Each approach has its strengths, and hybrid methods increasingly combine them to capture multiscale physiology.

Computational Fluid Dynamics

Patient-specific computational fluid dynamics (CFD) models reconstruct the pulmonary arterial tree from high-resolution computed tomography (CT) or magnetic resonance angiography (MRA). After segmenting the vessels, the Navier-Stokes equations are solved within the three-dimensional geometry, incorporating boundary conditions derived from catheterization or phase-contrast MRI. CFD provides detailed maps of velocity fields, wall shear stress, and pressure drops. For PH patients, simulations reveal flow disturbances—such as vortices at bifurcations and high oscillatory shear index—that correlate with disease severity and predict sites of future thrombosis. A landmark study by Kheyfets et al. (2018) demonstrated that CFD-derived parameters could distinguish PAH patients from controls with high sensitivity.

Lumped-Parameter and Zero-Dimensional Models

For clinical applications requiring rapid feedback, lumped-parameter (0D) models represent the entire pulmonary circulation as an electrical circuit, with resistors, capacitors, and inductors. These models simulate pressure-flow relationships using ordinary differential equations, making them computationally inexpensive and suitable for real-time decision support. When combined with patient-specific data such as PVR and compliance, 0D simulations can predict the hemodynamic effect of vasodilators or the impact of a balloon angioplasty on distal resistance. They are also used to estimate right ventricular–arterial coupling, a key predictor of prognosis in PH.

Patient-Specific Modeling Workflow

Creating a reliable simulation requires a multi-step pipeline: (1) image acquisition; (2) vessel segmentation and mesh generation; (3) assignment of material properties and boundary conditions; (4) numerical solution; (5) validation against clinical measurements. While labor-intensive, recent advances in deep learning have automated segmentation and mesh generation, reducing processing time from hours to minutes. Validation remains essential, and studies have shown that simulated pressures agree with invasive measurements within 2–4 mmHg in well-calibrated models.

Enhancing Diagnosis with Simulation

Simulation does not replace the right heart catheterization—the gold standard for PH diagnosis—but it adds complementary, spatially resolved information that can detect early disease and refine risk stratification.

Simulating Hemodynamic Parameters

Traditional diagnostic criteria rely on mPAP, PVR, and pulmonary artery wedge pressure. However, these global metrics may miss early vascular stiffening or focal obstruction. Simulation can compute regional compliance and resistance, identifying subtle abnormalities before they elevate global pressures. For example, a simulated decrease in distal arterial compliance may precede an increase in mPAP by months, offering a window for earlier intervention. Researchers have also developed surrogate markers—such as the ratio of pulse pressure to stroke volume derived from simulated waveforms—that correlate with disease severity and survival.

Identifying Subtle Flow Patterns

Vortex formation, flow recirculation, and high wall shear stress gradients are hallmarks of pathological hemodynamics. CFD simulations of PAH patients consistently show disrupted laminar flow, with vortex cores near the main pulmonary bifurcation. These features are not visible on standard angiography but can be extracted from simulations and have been linked to endothelial activation and disease progression. A 2020 study found that simulated flow pattern metrics improved classification of mild versus moderate PH beyond conventional echocardiographic indices.

Simulation in Treatment Planning

The most compelling promise of simulation lies in its ability to personalize treatment—allowing clinicians to test interventional strategies virtually before performing them on the patient.

Predicting Response to Vasodilators

In acute vasoreactivity testing, patients inhale nitric oxide or receive prostacyclin analogues while hemodynamics are measured. Simulation can extend this by modeling chronic drug effects. By altering vessel tone in a patient-specific model—reducing distal resistance or increasing compliance—simulations predict the long-term reduction in mPAP and right ventricular workload. This virtual drug testing could help identify non-responders earlier and reduce exposure to costly, side-effect-laden therapies.

Optimization of Percutaneous Interventions

Balloon pulmonary angioplasty (BPA) is an effective treatment for inoperable chronic thromboembolic pulmonary hypertension. However, determining which lesions to dilate and to what diameter remains operator-dependent. CFD simulations of BPA can compare multiple dilation strategies: for each simulated scenario, the algorithm computes the post-intervention pressure drop and flow distribution. Virtual BPA has been shown to identify the optimal lesion(s) to target, maximizing hemodynamic improvement while minimizing the risk of reperfusion edema. Published clinical series now use simulation to guide real procedures, and outcomes are being tracked in prospective registries.

Surgical Planning for Pulmonary Endarterectomy

Pulmonary endarterectomy (PEA) is the gold-standard treatment for proximal chronic thromboembolic disease. The success of surgery depends on complete removal of obstructive material, but residual PH occurs in 10–30% of cases. Preoperative CFD models can simulate the hemodynamic effect of removing specific thrombi, estimating the expected decrease in PVR and identifying segments where revascularization would yield the largest benefit. This virtual planning helps surgeons decide on the extent of dissection and may reduce the rate of residual PH.

Integrating Simulation into Clinical Workflow

Despite its potential, simulation remains largely a research tool. Widespread clinical adoption requires addressing hurdles related to automation, validation, and interoperability.

Automation and Real-Time Capability

Manual segmentation and mesh generation are too time-consuming for routine clinical use. Recent developments in convolutional neural networks enable fully automated segmentation of the pulmonary arteries from CTA in under a minute. Coupled with reduced-order models that run in seconds, “near-real-time” simulation is becoming feasible. Several academic groups and startups now offer cloud-based platforms where clinicians upload imaging data and receive simulated hemodynamic metrics within minutes.

Validation and Regulatory Approval

For simulation to guide therapy, its predictions must be rigorously validated against invasive measurements and patient outcomes. Multicenter studies have reported good accuracy for pressure and flow predictions (errors < 10%), but variability remains across centers and scanner protocols. Regulatory agencies like the FDA have begun to approve certain computational models as “medical devices” (e.g., for coronary flow reserve), and similar pathways are emerging for pulmonary applications. Once standardized, simulation could be reimbursed as a diagnostic adjunct.

Future Directions

The next decade will see simulation evolve from a specialized academic exercise to a routine component of pulmonary hypertension care.

Artificial Intelligence and Data Fusion

Machine learning can accelerate simulation by learning the mapping from imaging features to hemodynamic parameters, bypassing the need to solve fluid equations in real time. Hybrid models that combine physics-based simulation with data-driven corrections are particularly promising. Furthermore, incorporating wearable sensor data—such as continuous pulse oximetry or activity metrics—could enable “digital twin” simulations that update dynamically as the patient’s condition changes.

Integration with Imaging Advances

New imaging techniques, including 4D flow MRI and dual-energy CT, provide richer boundary conditions. 4D flow MRI directly measures velocity fields across the cardiac cycle, which can be used to validate and calibrate simulations. Dual-energy CT can map perfusion defects, allowing simulation to model not only arteries but also capillary-level flow. Combining these technologies will yield whole-lung models that simulate gas exchange and drug distribution.

Personalized Risk Stratification and Trials

Longitudinal simulations could track disease progression and predict the risk of clinical worsening. In clinical trials, simulated hemodynamic endpoints may serve as surrogate markers, reducing sample sizes and trial duration. The European Respiratory Society and other bodies are already incorporating computational modeling into guidelines for PH research.

Conclusion

Simulation of pulmonary circulation has matured from a theoretical pursuit into a practical tool that can improve the diagnosis and treatment of pulmonary hypertension. By providing a detailed, patient-specific view of hemodynamics, it enables earlier detection of disease, better risk stratification, and more precise interventional planning. While challenges remain in automation, validation, and clinical integration, the trajectory is clear: computational simulation will become an indispensable component of the pulmonary hypertension care pathway, ultimately translating into better outcomes for patients.