About

Noelia Grande Gutiérrez is an Assistant Professor in the Department of Mechanical Engineering at Carnegie Mellon University. Her research interests lie at the intersection of computational engineering and cardiovascular medicine, involving the development and application of multiphysics models that support clinical decision making and provide novel insights into cardiovascular disease. She has a strong academic background, having earned a doctorate in mechanical engineering from Stanford University in 2019, along with master's degrees in engineering sciences from UC San Diego and biomedical engineering from the University of Barcelona, as well as a bachelor's degree in aerospace engineering from the Technical University of Madrid. Grande Gutiérrez has been recognized with fellowships from the American Heart Association and la Caixa Foundation. She also served as a postdoctoral research fellow in the Diamond Lab at the University of Pennsylvania, where she worked on developing a multiscale computational framework to investigate patient-specific thrombosis in coronary artery disease patients, from initial platelet deposition to artery occlusion. Her work combines expertise in artificial intelligence, bioengineering, biomechanics, and computer-aided design, contributing to advancements in medical laboratory technology and cardiovascular research.

Research topics

• Medicine
• Cardiology
• Physics
• Internal medicine
• Mechanics
• Mathematics
• Optics
• Biomedical engineering
• Geometry
• Nanotechnology
• Materials science
• Biology
• Biophysics

Selected publications

• Scalable super-resolution of flow conveyance systems through adaptive domain decomposition

  Journal of Computing and Information Science in Engineering · 2026-05-06

  article

  Abstract Modern engineering design requires high-fidelity simulations, which can impose an enormous computational burden and slow the speed of design iteration. Data-driven up-sampling methods like physics-informed neural networks (PINNs) help reduce the computational resources required. However, machine learning model capacity and hardware limitations still pose challenges when evaluating large engineering simulations with complex physics dynamics. Recently, methods have been proposed to enforce the principle of locality in physical systems to neural network layers, allowing for concurrent inference on smaller subdomains with improved efficiency and accuracy. Based on such an idea, we extend the theory of domain decomposition to complex three-dimensional geometries using graph neural networks (GNNs). We developed a graph decomposition method to improve the training and inference efficiency of machine learning models. Super-resolution GNNs are then trained on individual subdomains distributed among GPU nodes. This approach significantly reduces computational overhead while maintaining simulation accuracy. We validate the method performance on two engineering applications: a variable inlet-angle mixing elbow junction and a low-pressure bleed duct from an Airbus A350 aircraft. These results demonstrate that our approach can effectively bridge the gap between computational efficiency and simulation fidelity across different scales of complex engineering design tasks.

  PublisherDOI

• Scale-dependent mechanisms of near-wall platelet enrichment in arterial flows

  Physics of Fluids · 2026-04-01

  articleOpen accessSenior author

  Platelet transport is essential for arterial thrombosis, yet the mechanisms governing near-wall platelet enrichment at arterial scales remain poorly understood. While red blood cell (RBC)–driven platelet margination is well established in microvascular flows, its relevance in larger arteries with higher Reynolds numbers and complex three-dimensional structures remains unclear. Here, we systematically examine platelet transport across arterial scales by comparing four platelet transport models, including inertial and non-inertial equations with and without a hematocrit-enhanced drift velocity. In axisymmetric flows, RBC-driven platelet margination induces a pronounced near-wall platelet enrichment at the arteriole scale, but this effect rapidly diminishes with increasing vessel diameter and Reynolds number, resulting in nearly uniform platelet distributions in arteries regardless of the transport equation employed. In contrast, simulations in three-dimensional curved vessels show that curvature-induced secondary flows generate substantial cross-stream platelet transport at arterial scales. The magnitude of this near-wall platelet enrichment correlates with the Dean number, indicating that secondary flows, rather than RBC-driven margination, dominate platelet transport in arteries. Together, these findings demonstrate that platelet transport mechanisms are scale-dependent: RBC-driven margination governs platelet transport in arterioles, whereas geometry-driven secondary flow advection governs arterial-scale transport. This work clarifies the physical relevance of hematocrit-enhanced drift models across vascular scales and informs the selection of platelet transport formulations for multiscale thrombosis simulations.

  PublisherDOI

• Computational Fluid Dynamics Analysis of Cerebrovascular Hemodynamic Differences in Adults with Sickle Cell Disease: Comparing Stroke and Non-Stroke Cohorts

  Annals of Biomedical Engineering · 2026-04-08

  articleOpen accessSenior authorCorresponding

  PURPOSE: Sickle cell disease (SCD) is a debilitating genetic disorder affecting hemoglobin in red blood cells. Patients with SCD are at risk of cerebrovascular disease in the Circle of Willis (CoW), with strokes occurring from early childhood into adulthood. Despite this lifelong risk, stroke prevention guidelines using transcranial Doppler ultrasound (TCD) exist only for children, leaving a critical gap for adults. This study aimed to characterize cerebral hemodynamics in the CoW of adults with SCD to support future risk stratification and treatment guidelines. METHODS: Numerical simulations were performed using 3D vascular geometries segmented from high-resolution, patient-specific magnetic resonance imaging in healthy controls (n=3), SCD patients without stroke (n=3), and SCD patients post-stroke (n=3). Key hemodynamic parameters including time-averaged wall shear stress (TAWSS), surface area exposed to low or high WSS, time-averaged mean of maximum velocity (TAMMV), and pressure drop across the CoW were quantified and compared. RESULTS: Pa. In contrast, SCD patients without stroke had the highest TAWSS and greatest area exposed to WSS >7 Pa. Despite similar total cerebral blood flow to controls, post-stroke patients showed a lower pressure drop across the CoW. CONCLUSIONS: Patient-specific simulations can quantify cerebral hemodynamics in adults with SCD, offering insight into stroke-related changes and informing future stroke risk assessment and personalized treatment strategies.

  PublisherOA PDFDOI

• Energy-based feature extraction with adaptive local domain decomposition for prediction of transient and turbulence flow with operator regression models

  Computers & Fluids · 2026-01-08

  articleSenior authorCorresponding
  PublisherDOI

• Fast Super-Resolution Analysis of Low-Pressure Duct Air Flow Through Adaptive Domain Decomposition

  2025-08-17

  article

  Abstract Modern engineering design requires high-fidelity simulations, which can impose an enormous computational burden and slow the speed of design iteration. Data-driven up-sampling methods like physics-informed neural networks (PINNs) help reduce the computational resources required. However, machine learning model capacity and hardware limitations still pose challenges when evaluating large engineering simulations with complex physics dynamics. Recently, methods have been proposed to enforce the principle of locality in physical systems to neural network layers, allowing for concurrent inference on smaller subdomains with improved efficiency and accuracy. Based on such an idea, we extend the theory of domain decomposition to complex three-dimensional geometries using graph neural networks (GNNs). We developed a graph decomposition method to improve the training and inference efficiency of machine learning models. Super-resolution GNNs are then trained on individual subdomains distributed among GPU nodes, and during the inference phase, their predictions are combined to achieve a close to linear time reduction as the number of parallel GPUs increases. This approach significantly reduces computational overhead while maintaining simulation accuracy. The parallel nature of our method allows for scalability across available hardware resources, making it suitable for industrial applications where time constraints are critical. We validate the method performance on the design and simulation of a low-pressure bleed duct from an Airbus A350 aircraft, achieving 0.9947 in R2 metric in velocity and 0.9996 in pressure compared with high-fidelity simulations. These results demonstrate that our approach can effectively bridge the gap between computational efficiency and simulation fidelity in complex engineering design tasks.

  PublisherDOI

• Sensitivity of coronary hemodynamics to vascular structure variations in health and disease

  Scientific Reports · 2025-01-27 · 6 citations

  articleOpen accessSenior author

  Local hemodynamics play an essential role in the initiation and progression of coronary artery disease. While vascular geometry alters local hemodynamics, the relationship between vascular structure and hemodynamics is poorly understood. Previous computational fluid dynamics (CFD) studies have explored how anatomy influences plaque-promoting hemodynamics. For example, areas exposed to low wall shear stress (ALWSS) can indicate regions of plaque growth. However, small sample sizes, idealized geometries, and simplified boundary conditions have limited their scope. We generated 230 synthetic models of left coronary arteries and simulated coronary hemodynamics with physiologically realistic boundary conditions. We measured the sensitivity of hemodynamic metrics to changes in bifurcation angles, positions, diameter ratios, tortuosity, and plaque topology. Our results suggest that the diameter ratio between left coronary branches plays a substantial role in generating adverse hemodynamic phenotypes and can amplify the effect of other geometric features such as bifurcation position and angle, and vessel tortuosity. Introducing mild plaque in the models did not change correlations between structure and hemodynamics. However, certain vascular structures can induce ALWSS at the trailing edge of the plaque. Our analysis demonstrates that coronary artery vascular structure can provide key insight into the hemodynamic environments conducive to plaque formation and growth.

  PublisherOA PDFDOI

• Enforcing the principle of locality for physical simulations with neural operators

  Journal of Computational Physics · 2025-06-04 · 2 citations

  articleOpen access

  • We show that deep learning permits violations of the principle of locality in time-dependent physical simulations. • We establish an efficient data decomposition method forcing neural operators to comply with the principle of locality. • Numerical experiments demonstrate that this method accelerates convergence and improves accuracy of benchmark neural operators. Time-dependent partial differential equations (PDEs) for classic physical systems are established based on the conservation of mass, momentum, and energy, which are ubiquitous in scientific and engineering applications. These PDEs are strictly local-dependent according to the principle of locality in physics, which means that the evolution at a point is only influenced by the neighborhood around it whose size is determined by the length of timestep multiplied with the speed of characteristic information traveling in the system. However, deep learning architecture cannot strictly enforce the local-dependency as it inevitably increases the scope of information to make local predictions as the number of layers increases. Under limited training data, the extra irrelevant information results in sluggish convergence and compromised generalizability. This paper aims to solve this problem by proposing a data decomposition method to strictly limit the scope of information for neural operators making local predictions, which is called data decomposition enforcing local-dependency (DDELD). The numerical experiments over multiple physical phenomena show that DDELD significantly accelerates training convergence and reduces test errors of benchmark models on large-scale engineering simulations.

  PublisherDOI

• Energy-Based Feature Extraction with Adaptive Local Domain Decomposition for Prediction of Transient and Turbulence Flow with Operator Regression Models

  SSRN Electronic Journal · 2025-01-01

  preprintOpen accessSenior author
  PublisherDOI

• Integrating local and distant radiation‐induced lung injury: Development and validation of a predictive model for ventilation loss

  Medical Physics · 2024-05-31 · 1 citations

  article

  BACKGROUND: Investigations on radiation-induced lung injury (RILI) have predominantly focused on local effects, primarily those associated with radiation damage to lung parenchyma. However, recent studies from our group and others have revealed that radiation-induced damage to branching serial structures such as airways and vessels may also have a substantial impact on post-radiotherapy (RT) lung function. Furthermore, recent results from multiple functional lung avoidance RT trials, although promising, have demonstrated only modest toxicity reduction, likely because they were primarily focused on dose avoidance to lung parenchyma. These observations emphasize the critical need for predictive dose-response models that effectively incorporate both local and distant RILI effects. PURPOSE: We develop and validate a predictive model for ventilation loss after lung RT. This model, referred to as P+A, integrates local (parenchyma [P]) and distant (central and peripheral airways [A]) radiation-induced damage, modeling partial (narrowing) and complete (collapse) obstruction of airways. METHODS: In an IRB-approved prospective study, pre-RT breath-hold CTs (BHCTs) and pre- and one-year post-RT 4DCTs were acquired from lung cancer patients treated with definitive RT. Up to 13 generations of airways were automatically segmented on the BHCTs using a research virtual bronchoscopy software. Ventilation maps derived from the 4DCT scans were utilized to quantify pre- and post-RT ventilation, serving, respectively, as input data and reference standard (RS) in model validation. To predict ventilation loss solely due to parenchymal damage (referred to as P model), we used a normal tissue complication probability (NTCP) model. Our model used this NTCP-based estimate and predicted additional loss due radiation-induced partial or complete occlusion of individual airways, applying fluid dynamics principles and a refined version of our previously developed airway radiosensitivity model. Predictions of post-RT ventilation were estimated in the sublobar volumes (SLVs) connected to the terminal airways. To validate the model, we conducted a k-fold cross-validation. Model parameters were optimized as the values that provided the lowest root mean square error (RMSE) between predicted post-RT ventilation and the RS for all SLVs in the training data. The performance of the P+A and the P models was evaluated by comparing their respective post-RT ventilation values with the RS predictions. Additional evaluation using various receiver operating characteristic (ROC) metrics was also performed. RESULTS: We extracted a dataset of 560 SLVs from four enrolled patients. Our results demonstrated that the P+A model consistently outperformed the P model, exhibiting RMSEs that were nearly half as low across all patients (13 ± 3 percentile for the P+A model vs. 24 ± 3 percentile for the P model on average). Notably, the P+A model aligned closely with the RS in ventilation loss distributions per lobe, particularly in regions exposed to doses ≥13.5 Gy. The ROC analysis further supported the superior performance of the P+A model compared to the P model in sensitivity (0.98 vs. 0.07), accuracy (0.87 vs. 0.25), and balanced predictions. CONCLUSIONS: These early findings indicate that airway damage is a crucial factor in RILI that should be included in dose-response modeling to enhance predictions of post-RT lung function.

  PublisherDOI

• Enforcing the Principle of Locality for Physical Simulations with Neural Operators

  arXiv (Cornell University) · 2024-05-02

  preprintOpen access

  Time-dependent partial differential equations (PDEs) for classic physical systems are established based on the conservation of mass, momentum, and energy, which are ubiquitous in scientific and engineering applications. These PDEs are strictly local-dependent according to the principle of locality in physics, which means that the evolution at a point is only influenced by the neighborhood around it whose size is determined by the length of timestep multiplied with the speed of characteristic information traveling in the system. However, deep learning architecture cannot strictly enforce the local-dependency as it inevitably increases the scope of information to make local predictions as the number of layers increases. Under limited training data, the extra irrelevant information results in sluggish convergence and compromised generalizability. This paper aims to solve this problem by proposing a data decomposition method to strictly limit the scope of information for neural operators making local predictions, which is called data decomposition enforcing local-dependency (DDELD). The numerical experiments over multiple physical phenomena show that DDELD significantly accelerates training convergence and reduces test errors of benchmark models on large-scale engineering simulations.

  PublisherOA PDFDOI

Frequent coauthors

• Alison L. Marsden

  11 shared

• Andrew M. Kahn

  University of California, San Diego

  9 shared

• Jane C. Burns

  Rady Children's Hospital-San Diego

  8 shared

• Wenzhuo Xu

  Shenzhen University

  7 shared

• Christopher McComb

  7 shared

• Mathew Mathew

  Philadelphia College of Osteopathic Medicine

  6 shared

• Brian W. McCrindle

  6 shared

• Galina Lyskina

  4 shared

Education

• B.S., Aerospace Engineering

  Technical University of Madrid

• M.A., Biomedical Sciences

  University of Barcelona

• M.S., Engineering Sciences

  UC San Diego

• Ph.D., Mechanical Engineering

  Stanford University

  2019

Awards & honors

• Fellowship from the American Heart Association
• Fellowship from la Caixa Foundation