Quaderni MOX
Pubblicazioni
del Laboratorio di Modellistica e Calcolo Scientifico MOX. I lavori riguardano prevalentemente il campo dell'analisi numerica, della statistica e della modellistica matematica applicata a problemi di interesse ingegneristico. Il sito del Laboratorio MOX è raggiungibile
all'indirizzo mox.polimi.it
Trovati 1349 prodotti
-
41/2020 - 07/06/2020
Cannistrà,M.; Masci, C.; Ieva, F.; Agasisti, T.; Paganoni, A.M.
Not the magic algorithm: modelling and early-predicting students dropout through machine learning and multilevel approach | Abstract | | According to OECD, almost 30 per cent of students leave tertiary education programs without obtaining
a degree. This number measures a dead loss of human capital and a waste of public and private
resources. This paper contributes to the existing knowledge about students dropout by combining a
theoretical-based model with a data-driven approach to detect students who are more likely to leave
university in the first year. We propose the use of multilevel statistical models and machine learning
methods, applied to administrative data from a leading Italian university. The findings are encouraging,
as the methodology is able to predict at-risk students very precisely. We provide evidence of
the essential role of data relative to early performance (i.e. grades obtained in the first semester).
Moreover, the selection of major strongly influences the probability of dropping out. |
-
40/2020 - 07/06/2020
Fresca, S.; Manzoni, A.; Dedè, L.; Quarteroni, A.
Deep learning-based reduced order models in cardiac electrophysiology | Abstract | | Predicting the electrical behavior of the heart, from the cellular scale to the tissue level, relies on the formulation and numerical approximation of coupled nonlinear dynamical systems. These systems describe the cardiac action potential, that is the polarization/depolarization cycle occurring at every heart beat that models the time evolution of the electrical potential across the cell membrane, as well as a set of ionic variables. Multiple solutions of these systems, corresponding to different model inputs, are required to evaluate outputs of clinical interest, such as activation maps and action potential duration. More importantly, these models feature coherent structures that propagate over time, such as wavefronts. These systems can hardly be reduced to lower dimensional problems by conventional reduced order models (ROMs) such as, e.g., the reduced basis (RB) method. This is primarily due to the low regularity of the solution manifold (with respect to the problem parameters) as well as to the nonlinear nature of the input-output maps that we intend to reconstruct numerically. To overcome this difficulty, in this paper we propose a new, nonlinear approach which exploits deep learning (DL) algorithms to obtain accurate and efficient ROMs, whose dimensionality matches the number of system parameters. Our DL approach combines deep feedforward neural networks (NNs) and convolutional autoencoders (AEs). We show that the proposed DL-ROM framework can efficiently provide solutions to parametrized electrophysiology problems, thus enabling multi-scenario analysis in pathological cases. We investigate three challenging test cases in cardiac electrophysiology and prove that DL-ROM outperforms classical projection-based ROMs. |
-
39/2020 - 07/06/2020
Martinolli, M.; Biasetti, J.; Zonca, S.; Polverelli, L.; Vergara, C.
Extended Finite Element Method for Fluid-Structure Interaction in Wave Membrane Blood Pumps | Abstract | | Numerical simulations of cardiac blood pump systems are integral to the optimization of device design, hydraulic performance and hemocompatibility. In wave membrane blood pumps, blood propulsion arises from the wave propagation along an oscillating immersed membrane, which generates small pockets of fluid that are pushed towards the outlet against an adverse pressure gradient.
We studied the Fluid-Structure Interaction between the oscillating membrane and the blood flow via three-dimensional simulations using the Extended Finite Element Method, an unfitted numerical technique that avoids remeshing by using a fluid fixed mesh.
Our three-dimensional numerical simulations in a realistic pump geometry highlighted the role of the membrane deformation in promoting a blood flow towards the outlet despite of a resistive pressure gradient. We also simulated the pump system at different pressure conditions and we validated the numerical results against textit{in-vitro} experimental data. |
-
38/2020 - 03/06/2020
Sollini, M.; Kirienko, M.; Cavinato, L.; Ricci, F.; Biroli, M.; Ieva, F.; Calderoni, L.; Tabacchi, E.; Nanni, C.; Zinzani, P.L.; Fanti, S.; Guidetti, A; Alessi, A.; Corradini, P.; Seregni, E.; Carlo-Stella, C.; Chiti, A.
Methodological framework for radiomics applications in Hodgkin’s lymphoma | Abstract | | Background: According to published data, radiomics features differ between lesions of refractory/relapsing HL patients from those of long-term responders. However,several methodological aspects have not been elucidated yet.
Purpose: The study aimed at setting up a methodological framework in radiomics applications in Hodgkin’s lymphoma (HL), especially at (a) developing a novel feature selection approach, (b) evaluating radiomic intra-patient lesions’ similarity, and (c) classifying relapsing refractory (R/R) vs non-(R/R) patients.
Methods: We retrospectively included 85 patients (male:female = 52:33; median age 35 years, range 19–74). LIFEx (www.lifexsoft.org) was used for [18F]FDG-PET/CT segmentation and feature extraction. Features were a-priori selected if they were highly correlated or uncorrelated to the volume. Principal component analysis transformed features were used to build the fingerprints that were tested to assess lesions’ similarity, using the silhouette. For intra-patient similarity analysis, we used patients having multiple lesions only. To classify patients as non-R/R and R/R, the fingerprint considering one single lesion (fingerprint_One) and all lesions (fingerprint_All) was tested using Random Undersampling Boosting of Tree Ensemble (RUBTE).
Results: HL fingerprints included up to 15 features. Intra-patient lesion similarity analysis resulted in mean/median silhouette values below 0.5 (low similarity especially in the non-R/R group). In the test set, the fingerprint_One classification accuracy was 62% (78% sensitivity and 53% specificity); the classification by RUBTE using fingerprint_All resulted in 82% accuracy (70% sensitivity and 88% specificity).
Conclusions: Lesion similarity analysis was developed, and it allowed to demonstrate that HL lesions were not homogeneous within patients in terms of radiomics signature. Therefore, a random target lesion selection should not be adopted for radiomics applications. Moreover, the classifier to predict R/R vs non-R/R performed the best when all the lesions were used. |
-
37/2020 - 24/05/2020
Fumagalli, A.; Scotti, A.
A mathematical model for thermal single-phase flow and reactive transport in fractured porous media | Abstract | | In this paper we present a mathematical model and a numerical workflow for the simulation of a thermal single-phase flow with reactive transport in porous media, in the presence of fractures. The latter are thin regions which might behave as high or low permeability channels depending on their physical parameters, and are thus of paramount importance in underground flow problems. Chemical reactions may alter the local properties of the porous media as well as the fracture walls, changing the flow path and possibly occluding some portions of the fractures or zones in the porous media. To solve numerically the coupled problem we propose a temporal splitting scheme so that the equations describing each physical process are solved sequentially. Numerical tests shows the accuracy of the proposed model and the ability to capture complex phenomena, where one or multiple fractures are present. |
-
36/2020 - 24/05/2020
Pellagatti, M.; Masci, C.; Ieva, F.; Paganoni A.M.
Generalized Mixed-Effects Random Forest: a flexible approach to predict university student dropout | Abstract | | We propose a new statistical method, called Generalized Mixed-Effects Random Forest (GMERF), that extends the use of random forest to the analysis of hierarchical data, for any type of response variable in the exponential family, considering both continuous and discrete covariates and without assuming a closed form in the association between the response and the fixed-effects covariates. At the same time GMERF takes into consideration the nested structure of hierarchical data, modelling
the latent grouping structure that exists in the higher level of the hierarchy and allowing statistical inference on this structure. In the case study, we apply GMERF to Higher Education data to analyse the university students dropout; in particular, we are interested in predicting students dropout probability given students-level information and considering the degree program they are enrolled in as the grouping factor. |
-
35/2020 - 24/05/2020
Morbiducci, U.; Mazzi, V.; Domanin, M.; De Nisco, G.; Vergara, C.; Steinman, D.A.; Gallo, D.
Wall shear stress topological skeleton independently predicts long-term restenosis after carotid bifurcation endarterectomy | Abstract | | Wall shear stress (WSS) topological skeleton, composed by fixed points and the manifolds linking them, reflects the presence of blood flow features associated to adverse vascular response. However, the influence of WSS topological skeleton on vascular pathophysiology is still underexplored. This study aimed to identify direct associations between the WSS topological skeleton and markers of vascular disease from real-world clinical longitudinal data of long-term restenosis after carotid endarterectomy (CEA).
Personalized computational hemodynamic simulations were performed on a cohort of 13 carotid models pre-CEA and at 1 month after CEA. At 60 months after CEA, intima-media thickness (IMT) was measured to detect long-term restenosis. The analysis of the WSS topological skeleton was carried out by applying a Eulerian method based on the WSS vector field divergence. To provide objective thresholds for WSS topological skeleton quantitative analysis, a computational hemodynamic dataset of 46 ostensibly healthy carotid bifurcation models was considered.
CEA interventions did not completely restore physiological WSS topological skeleton features. Significant associations emerged between IMT at 60 months follow-up and the exposure to (1) high temporal variation of WSS contraction/expansion (R2=0.51, p<0.05), and (2) high fixed point residence times, weighted by WSS contraction/expansion strength (R2=0.53, p<0.05). These WSS topological skeleton features were statistically independent from the exposure to low WSS, a previously reported predictor of long-term restenosis, therefore representing different hemodynamic stimuli and potentially impacting differently the vascular response. This study confirms the direct association between WSS topological skeleton and vascular response, contributing to elucidate the mechanistic link between flow disturbances and clinical observations of vascular lesions.
|
-
34/2020 - 24/05/2020
Antonietti, P.F.; Botti, M.; Mazzieri, I.; Nati Poltri, S.
A high-order discontinuous Galerkin method for the poro-elasto-acoustic problem on polygonal and polyhedral grids | Abstract | |
The aim of this work is to introduce a discretization of the physical phenomenon of propagation of acoustic waves through poroelastic materials, by exerting a finite element discontinuous Galerkin method on polygonal meshes. Wave propagation is modeled by the acoustics equations in the acoustic domain and the low-frequency Biot’s equations in the poroelastic one. The coupling is introduced by considering (physically consistent) interface conditions, imposed on the interface between the domains, modelling both open and sealed pores. Existence and uniqueness is proven for the strong formulation based on employing the Hille-Yosida theorem. For the space discretization we introduce a discontinuous Galerkin method, which is then coupled with suitable time integration schemes, such as the leapfrog and the Newmark methods. A stability analysis both for the continuous problem and the semi-discrete one is presented and error estimates for the energy norm are derived. A wide set of numerical results obtained on test cases with manufactured solutions are presented in order to validate the error analysis. Examples of physical interest are also presented to test the capability of the proposed methods in practical cases.
|
|
