arXiv subjects

Federica Ninno

Publications and source records attributed to Federica Ninno.

3 recordsLinked to original sources

arXiv2026

Physics-Consistent Neural Networks for Learning Deformation and Director Fields in Microstructured Media with Loss-Based Validation Criteria

In this work, we study the mechanical behavior of solids with microstructure using the framework of Cosserat elasticity with a single unit director. This formulation captures the coupling between deformation and orientational fields that arises in many structured materials. To compute equilibrium configurations of such media, we develop two complementary computational approaches: a finite element formulation based on variational principles and a neural network-based solver that directly minimizes the total potential energy. The neural architecture is constructed to respect the fundamental kinematic structure of the theory. In particular, it enforces frame invariance of the energy, satisfies the unit-length constraint on the director field, and represents deformation and director fields through separate networks to preserve their kinematic independence in the variational setting. Beyond satisfying balance laws, however, physically admissible solutions must also correspond to stable energy minimizers. To assess this requirement, we derive the quasiconvexity condition, rank-one convexity condition, and the Legendre-Hadamard inequalities for the Cosserat model and formulate them in a manner suitable for evaluating neural network predictions. These necessary stability conditions provide a physics-based validation framework: network outputs that violate these necessary conditions cannot correspond to stable energy minimizers and can therefore be rejected. In this way, we integrate classical variational stability theory with modern machine-learning solvers, establishing a computational workflow in which equilibrium solutions are not only learned but also assessed for energetic consistency.

cs.LG↗

arXiv2025

A data-driven framework to identify restenosis-prone regions in femoral arteries from geometric and inflow waveform parameters

Haemodynamic indices derived from Computational Fluid Dynamics (CFD), such as Time-averaged Wall Shear Stress (TAWSS) and Oscillatory Shear Index (OSI), are closely associated with restenosis risk in Peripheral Arterial Disease (PAD). However, translating these insights into clinical practice may require computationally efficient approaches such as Reduced Order Model (ROM) or Machine Learning (ML). In this work, we developed an ML-ROM framework to predict critical, restenosis-prone, haemodynamic regions accounting for both vessel geometries and inlet flow waveforms. We generated 500 synthetic femoral-artery geometries parameterised by six geometric parameters, and created physiologically realistic inflow waveforms via Principal Component Analysis (PCA) of patient data. CFD was used to obtain the Wall Shear Stress (WSS) distribution, from which TAWSS and OSI were computed. Critical regions were then defined by applying threshold-based criteria to the TAWSS and OSI. Four critical-region definitions were considered: two with vessel-specific relative thresholds (TAWSS< 33rd percentile and OSI> 66nd percentile) and two with absolute thresholds (TAWSS< 0.5 Pa and OSI> 0.2). Proper orthogonal decomposition (POD) was then applied to these high-dimensional critical-region data to obtain ROMs; These were then used to train ML models from which the critical region regions could be reconstructed. Three ML architectures were explored: a Fourier-based architecture, a Long Short-term Memory (LSTM) architecture, and a Convolutional Neural Network (CNN) architecture. The Fourier models achieved the highest performance, with the median values of Balanced Accuracy (BA) exceeding 0.92 across all critical-region definitions. The ML-ROM framework also offered a substantial speed-up ratio, about nine orders of magnitude faster than traditional CFD.

physics.flu-dyn↗

arXiv2024

ML-ROM Wall Shear Stress Prediction in Patient-Specific Vascular Pathologies under a Limited Clinical Training Data Regime

High-fidelity numerical simulations such as Computational Fluid Dynamics (CFD) have been proven effective in analysing haemodynamics, offering insight into many vascular conditions. However, these methods often face challenges of high computational cost and long processing times. Data-driven approaches such as Reduced Order Modeling (ROM) and Machine Learning (ML) are increasingly being explored alongside CFD to advance biomechanical research and application. This study presents an integration of Proper Orthogonal Decomposition (POD)-based ROM with neural network-based ML models to predict Wall Shear Stress (WSS) in patient-specific vascular pathologies. CFD was used to generate WSS data, followed by POD to construct the ROM. The ML models were trained to predict the ROM coefficients from the inlet flowrate waveform, which can be routinely collected in the clinic. Two ML models were explored: a simpler flowrate-coefficients mapping model and a more advanced autoregressive model. Both models were tested against two case studies: flow in Peripheral Arterial Disease (PAD) and flow in Aortic Dissection (AD). Despite the limited training data sets (three flowrate waveforms for the PAD case and two for the AD case), the models were able to predict the haemodynamic indices, with the flowrate-coefficients mapping model outperforming the autoregressive model in both case studies. The accuracy is higher in the PAD case study, with reduced accuracy in the more complex case study of AD. Additionally, the computational cost analysis reveals a significant reduction in computational demands, with speed-up ratios on the order of $10^4$ for both case studies. This approach shows an effective integration of ROM and ML techniques for fast and reliable evaluations of haemodynamic properties that contribute to vascular conditions, setting the stage for clinical translation.

physics.flu-dyn↗