The clinical backdrop is sobering. Ascending thoracic aortic aneurysms, marked by abnormal enlargement of the aortic lumen, affect an estimated 5.3 per 100,000 people annually and arise from progressive degeneration that weakens the arterial wall. Surgical guidelines recommend replacing the ascending aorta once its diameter exceeds 5.5 centimeters in men and 5 centimeters in women, because enlargement raises the risk of rupture or dissection. The replacement graft, typically made of Dacron, is prized for its biocompatibility and durability, but it is far stiffer and less compliant than living aortic tissue. Surgeons have long reported post-operative complications, including type-B dissection of the descending aorta, distal to the implanted graft, with the aortic isthmus — the narrow segment just beyond the origin of the left subclavian artery, where the arch meets the descending aorta — recognized as the most vulnerable site. What has remained unclear is whether the increased risk stems from altered hemodynamic loading caused by the stiff proximal graft, or simply from the fact that extending a patient’s lifespan gives more time for other regions to deteriorate.

The new study does not answer that question definitively, but it equips the field with the tool needed to try. The research team, led by Ione Ianniruberto, Davide Astori, Emiliano Votta, and Alberto Redaelli at Politecnico di Milano together with collaborators at Weill Cornell Medicine, Yale University, and elsewhere, developed a fully automated patient-specific fluid–structure interaction (FSI) pipeline. FSI simulation is the computational gold standard for capturing how flowing blood deforms a vessel wall and how, in turn, the moving wall reshapes the flow. Earlier attempts by the same group and others relied on simplified surrogate approaches that treated the aortic wall as a thin linear elastic membrane, or on fully coupled simulations that were so labor-intensive they could only be applied to a single patient. The new framework, implemented in the open-source software SimVascular, automates nearly every step, transforming what once took weeks of manual model building into a standardized, repeatable workflow.

The machinery behind the pipeline is a carefully orchestrated marriage of clinical imaging and computational mechanics. High-resolution magnetic resonance angiography provides the three-dimensional anatomy of each patient’s thoracic aorta, which is automatically segmented using an artificial-intelligence-based tool called TotalSegmentator and converted into a luminal surface model. Cine-MRI, a functional imaging modality that captures the aorta pulsing through the cardiac cycle in forty frames, supplies regional measurements of wall thickness and systole-to-diastole area change — data that encode how much the vessel expands with each heartbeat. Four-dimensional flow MRI, which measures the full velocity vector of blood throughout the aortic volume at every point in the cardiac cycle, provides the inlet velocity profiles that drive the simulations. The algorithm even corrects the raw 4D flow data for eddy-current-induced errors and aliasing artifacts, and coregisters it with the anatomy using a computed phase-contrast angiography image, ensuring that velocities land in the right place in the reconstructed geometry.

Once the imaging data are harvested, the pipeline assigns mechanical properties to the aortic wall through an approach the authors describe as “patient-informed.” Because living tissue cannot simply be pulled apart to measure its stiffness before surgery, the team built a database of more than eighty candidate stress–strain curves drawn from published equi-biaxial tensile tests of human aortic tissue, classified by patient age group and underlying etiology — whether the aneurysm is degenerative or heritable, as in Marfan and Loeys–Dietz syndromes. For each patient, the algorithm selects the literature curve that best reproduces the area change actually measured from that patient’s Cine-MRI, using the Laplace equation to estimate circumferential wall stress from pressure, radius, and thickness. The selected curves are then fitted to the incompressible, hyperelastic, anisotropic Holzapfel–Gasser–Ogden constitutive model, which mathematically describes how two families of collagen fibers embedded in the aortic wall resist stretching — a crucial refinement over earlier isotropic models that ignored the tissue’s directional architecture.

Blood itself is modeled as an incompressible Newtonian fluid with a density of 1060 kilograms per cubic meter and a viscosity of 4 centipoise, a standard and well-justified simplification for the high-shear environment of large arteries. At the aortic inlet, the measured velocity field is applied directly, with velocity forced to zero along the wall edges to satisfy the no-slip condition. At the outlets — the three supra-aortic branches and the descending aorta — the pipeline attaches lumped-parameter Windkessel models, electrical-circuit-like representations of the downstream vasculature whose resistances and compliances are automatically tuned by an optimization algorithm until the simulated systolic and diastolic pressures match the patient’s measured brachial blood pressure and the simulated descending-aortic flow matches the 4D flow MRI waveform. The surrounding tissue’s mechanical support is captured through a Robin boundary condition with an external stiffness of 10,000 pascals per millimeter.

The full simulation sequence unfolds in three stages. First, a computational fluid dynamics simulation on a rigid-wall model computes the pressure field acting on the luminal surface at mean arterial pressure. Second, a structural analysis applies that pressure to the aortic wall to estimate its prestress state — the internal tension the vessel carries even at rest, which must be known to initialize realistic FSI simulations. Third, the fully coupled FSI simulation brings fluid and solid together, initialized with the computed pressure and prestress, and runs two complete cardiac cycles with a time step of just 0.1 milliseconds. The team applied this workflow to five patients with ascending aortic aneurysms treated at New York-Presbyterian Hospital between October 2023 and May 2025, each imaged within a month before surgery and again within six months after, yielding ten patient-specific simulations that systematically compared the diseased pre-operative state against the grafted post-operative one.

The results, while preliminary in a cohort of five, are remarkably consistent — and they land precisely where clinicians worry most. After graft implantation, the simulations showed increased flow-induced wall shear stress, the frictional force that blood exerts on the endothelial lining, along with increases in the time-averaged wall shear stress (TAWSS) and in the oscillatory shear index (OSI), a measure of how chaotically the direction of wall friction reverses over the cardiac cycle. All three indices rose consistently in the aortic isthmus, the region most commonly associated with post-operative dilation. Because endothelial cells are exquisitely sensitive sensors of shear stress, sustained elevation and oscillation of these forces are known triggers of adverse biological responses, including inflammatory signaling and structural degeneration of the vessel wall. In effect, the simulations suggest that stiffening the proximal aorta may redirect hemodynamic stress onto the one segment least equipped to handle it.

The study went further than scalar indices, adding a topological analysis of the wall shear stress field that tracked so-called fixed points — locations where the surface shear pattern has critical points, akin to stagnation zones in the flow landscape. Before surgery, these fixed points clustered in the ascending aorta; after grafting, they redistributed toward the isthmus region, a migration the authors interpret as a signature of localized hemodynamic disturbance that may correlate with adverse remodeling. The statistical framework was deliberately conservative: field variables were sampled in percentile distributions across a standardized isthmus region of interest, verified for non-normality with a Shapiro–Wilk test, and compared with paired non-parametric Wilcoxon signed-rank tests, with a ninety-ninth percentile threshold used to quantify near-maximum values while ruling out numerical artifacts.

What makes the work resonate beyond the aortic community is its automation. Previous state-of-the-art FSI studies validated their frameworks on single patients precisely because model generation was too manual to scale. By contrast, this pipeline — from DICOM images to simulation-ready input files — runs end-to-end through a single Python script, opening the door to longitudinal studies that could follow large cohorts of grafted patients and correlate early hemodynamic changes with later aortic remodeling. The authors are candid that five patients constitute a preliminary cohort, and the framework still depends on literature-derived stress–strain curves rather than direct tissue measurements, but the infrastructure now exists to test the compliance-mismatch hypothesis at scale. If larger studies confirm that rigid grafts systematically elevate shear stress in the isthmus, the implication for device engineering is direct: next-generation aortic grafts should be designed with controlled compliance to soften the hemodynamic impact downstream — turning a simulation pipeline into a blueprint for the operating room.

A new systematic review has charted how physics-informed neural networks are being used to simulate blood moving through narrowed arteries, highlighting both the promise of artificial intelligence for cardiovascular modeling and the obstacles that still stand between laboratory algorithms and clinical tools. The review examines research published from 2019 through November 2025 on simplified two-dimensional artery models containing localized stenosis, the technical term for a constriction caused by the buildup or deformation of material within a blood vessel. Such narrowed regions can accelerate flow, alter pressure, generate recirculation zones and expose vessel walls to abnormal mechanical forces. By embedding the governing laws of fluid mechanics directly into the training of neural networks, researchers are attempting to reproduce these changes with less computational cost than conventional numerical simulation. The review, published in Neural Computing and Applications, identifies a rapidly expanding field in which machine learning is being combined with computational fluid dynamics rather than simply replacing it.

The study by Sunday Akinwamide, Farhan Mohamed, Mohd Shahrizal Sunar and colleagues follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses, or PRISMA 2020, framework. The authors searched eight databases and initially identified 324 records. After screening and applying their inclusion criteria, 32 studies remained for detailed analysis. The selected papers focused specifically on blood-flow modeling in simplified two-dimensional stenosed arteries using synthetic datasets, allowing the review team to compare how different investigators formulated their equations, represented artery geometry, imposed boundary conditions, generated training data and evaluated model performance. This narrow scope was deliberate. Although physics-informed neural networks are now being applied to aneurysms, deformable vessels, brain circulation and three-dimensional vascular trees, simple two-dimensional channels provide a controlled test bed. They make it easier to determine whether an algorithm is learning the underlying physics or merely fitting a particular set of simulated examples.

Traditional computational fluid dynamics remains the reference method for many blood-flow simulations. In a typical CFD calculation, the artery is divided into a mesh of small elements, and numerical solvers approximate the Navier–Stokes equations throughout that domain. These equations express conservation of momentum, while the continuity equation enforces conservation of mass. For an incompressible fluid, continuity requires the velocity field to have zero divergence, meaning that fluid cannot mysteriously appear or disappear inside the vessel. The momentum equations account for pressure, viscosity and inertial effects. Blood is often approximated as a Newtonian fluid in basic models, although its real rheology is more complicated: it contains cells, changes viscosity with shear rate and can behave as a non-Newtonian fluid under some conditions. CFD can resolve these details with high accuracy, but the result may depend strongly on mesh quality. A constriction creates steep velocity and pressure gradients, requiring fine meshes and substantial computing time.

A PINN approaches the same problem differently. Instead of relying entirely on a mesh, it uses a neural network to represent unknown quantities such as velocity and pressure as continuous functions of spatial coordinates and, when needed, time. During training, the network is penalized not only when its predictions disagree with available data but also when those predictions violate physical laws. A typical loss function may combine residuals from the continuity equation and momentum equations with errors at the inlet, outlet and vessel walls. Automatic differentiation calculates the derivatives needed for these residuals directly from the network, allowing the model to evaluate whether its predicted velocity and pressure fields satisfy the equations at selected points throughout the artery. Boundary conditions can specify an inlet velocity or pressure waveform, an outlet condition and a no-slip wall, in which fluid immediately adjacent to a stationary vessel wall has zero relative velocity. The network therefore learns a solution constrained by both data and mechanics.

This design can be especially useful when measurements are sparse. In a clinical setting, imaging may reveal the shape of an artery while providing limited information about the full velocity field or pressure distribution. A purely data-driven model could produce plausible-looking results while violating conservation laws or predicting physically impossible flow. A PINN can use the available observations together with governing equations to fill in missing information. In principle, the same framework can be used for forward problems, in which flow is predicted from known vessel geometry and conditions, and inverse problems, in which unknown quantities such as pressure, viscosity or boundary parameters are inferred from partial observations. The review describes this data-efficient and potentially generalizable character as one of the main reasons PINNs have attracted interest in hemodynamics. However, data efficiency does not mean that training is automatically easy or that sparse input guarantees clinical accuracy.

The reviewed studies reveal several recurring patterns in how researchers have adapted PINNs to stenosed arteries. Many use synthetic datasets produced by analytical solutions or established CFD solvers. These datasets allow researchers to know the “correct” velocity and pressure fields against which neural-network predictions can be compared. Others use hybrid PINN–CFD frameworks, in which conventional numerical methods supply high-quality information while the neural network accelerates repeated calculations, reconstructs fields or handles selected parts of the solution. The review reports a steady rise in such hybrid approaches. This trend reflects a practical shift in the field: rather than presenting artificial intelligence as a wholesale substitute for physics-based simulation, researchers are using it as a surrogate, correction model or computational companion. Mesh-free formulations are also becoming more prominent because they can avoid some of the difficulties associated with generating and refining meshes around irregular or sharply narrowed geometries.

The geometry of stenosis is a critical source of complexity. In a straight two-dimensional artery, a localized narrowing reduces the cross-sectional area available to the fluid. For a given volumetric flow rate, the average velocity must increase through the constricted section. The pressure field responds to the changing geometry, and downstream flow may separate from the wall, creating regions of recirculation. These effects can be described using quantities such as the Reynolds number, which compares inertial and viscous forces, and wall shear stress, or WSS, which measures the tangential force exerted by flowing blood on the vessel wall. WSS is calculated from the near-wall velocity gradient and blood viscosity. It is biologically important because endothelial cells respond to mechanical stimuli, and disturbed or oscillatory flow has been associated with vascular disease processes. Yet the review identifies WSS estimation as one of the most persistent weaknesses in current PINN studies. Small errors in velocity gradients near the wall can produce much larger errors in calculated shear stress, even when the overall flow field appears accurate.

Another challenge is the enforcement of boundary conditions. The performance of a PINN depends on how the loss function balances the interior physics equations against constraints at the inlet, outlet and walls. If the network focuses too heavily on matching boundary data, it may leave substantial equation residuals inside the domain. If it prioritizes the governing equations, it may satisfy them while producing inaccurate inlet or wall behavior. The problem becomes more difficult for pulsatile blood flow, where the inlet condition changes over time, and for models that attempt to represent elastic arterial walls or fluid–structure interaction. The review notes that researchers have experimented with adaptive weighting, progressive boundary complexity and other training strategies to address these imbalances. The neural network’s architecture, the location and density of collocation points, the choice of activation functions and the optimization schedule can all influence convergence. These are not minor implementation details: two models using the same physical equations can produce very different results if their training procedures sample the domain or weight the loss terms differently.

Scalability remains a central concern. PINNs can reduce the cost of evaluating a trained surrogate, but training may itself be computationally demanding, particularly when the model must represent sharp gradients, multiple flow regimes or many combinations of geometry and boundary conditions. A network trained for one stenosis shape may not generalize reliably to another, and a model designed for steady laminar flow may struggle with pulsatile or turbulent-like conditions. The review also finds that studies do not yet use consistent benchmarks. Researchers may report different error metrics, sample different regions of the artery or compare their networks with different CFD references. Without standardized geometries, physical parameters, boundary conditions and evaluation protocols, it is difficult to determine whether one method is genuinely more accurate or simply tested under more favorable circumstances. The authors argue that reproducible benchmarking is essential if the field is to progress from promising demonstrations to dependable hemodynamic analysis.

The review’s broader conclusion is cautiously optimistic. PINNs provide a framework for linking computational physics, biomedical engineering and machine learning, and they may eventually support faster simulations, parameter estimation and personalized vascular modeling. Their ability to incorporate physical laws could be valuable when patient-specific measurements are incomplete, while hybrid systems may preserve the reliability of CFD and reduce the burden of repeated calculations. But the evidence does not justify treating these models as ready-made diagnostic instruments. The studies considered in the review largely rely on simplified two-dimensional geometries and synthetic data rather than the full anatomical complexity and measurement uncertainty found in patients. Before clinical translation, researchers will need stronger validation against experiments, medical imaging and established numerical solvers; more reliable near-wall predictions; clearer treatment of blood’s non-Newtonian behavior and vessel elasticity; and common standards for reporting error and uncertainty. For now, the significance of the work is less that an artificial neural network has solved blood flow than that the field is beginning to define the conditions under which such a solution can be trusted.