Titles:

AL-ALI Safaa  "Bayesian inference for cardiac electrophysiology with device tissue interactions for Artificial Pacemaker simulation using sheep data

CHAPOTTE Dylan "Bayesian Inference for the Electrical Impedance Tomography Problem

CLAIRON Quentin "Variational autoencoder for inference of nonlinear mixed effect models based on ordinary differential equations

FOSTIER Louis "PINNs for size-structured population dynamics inference

FOURCADE Sylvain "Incorporating measurement noise in EIT inverse problem"

GOSSARD Audrey "Parameter Estimation of Membrane Properties After Electroporation Using PDE-Based Impedance Modeling"

HAMAD Nour "An Anatomically Realistic Thoracic Conductivity Library for Bayesian EIT Inverse Problems"

MORANGE Martin "A multiscale Luenberger observer based on Wasserstein metric for collective motion models"

ZAYENI Hatem "Fading Regularization for Data Assimilation in Cell-Substrate Mechanical Interactions"

 

CHAPELLE Dominique

Title: Multiscale biomechanical modeling of the heart for clinical digital twins

Abstract: 

Cardiac contraction originates at the sub-cellular level within specific components of the cardiomyocytes (i.e. cardiac cells) called sarcomeres. This contractile behaviour then needs to be integrated at the organ level, namely, with a specific complex structure and shape. Furthermore, this organ crucially interacts with other physiological systems, the first of which being blood circulation via the cardiac function itself, and such ineractions must be adequately represented in order to obtain accurate and predictive model simulations. Once the modeling components have been thorougly validated, they can be used in combination with patient-specific data to build digital twins, i.e. personalized models that can be queried over time to predict pathological evolutions, or the effect of various possible treatments. This presentation will provide an overview of recent advances on cardiac modeling achieved in the ANANKE group (Inria and Ecole Polytechnique, with a particular focus on the key multiscale and integrated system modeling aspects that need to be addressed, and with many associated challenges pertaining to cardiovascular digital twin applications in cardiology and critical care, in particular.


SCHRODER Jesper

Title: Energy optimal observers -- a deterministic approach to state estimation

Abstract:
State estimation for evolving dynamical systems subject to noise is an increasingly important task in biomedicine and the life sciences. The Mortensen observer is a particular approach addressing this challenge. In contrast to stochastic filtering, where noise is understood as random processes, this approach models the disturbances as deterministic but unknown. At each time instant, an estimate is constructed by driving the model with minimum-energy disturbances that match the model and the measured output. The sequential characterization of the observer relies on the associated value function, a tool from optimal control theory solving the Hamilton-Jacobi-Bellman (H=JB) equation.
In this talk, we present an analytically rigorous foundation for the Morten sen observer applied to the class of quadratic ODEs and to a semilinear wave equation. Further, we propose numerical schemes for its approximation that avoid the computational complexity of directly solving the HJB equation. Numerical results show improved accuracy over the extended Kalman filter. Finally, we discuss directions for extending the concept to more challenging settings, such as parabolic systems or systems with modelun certainties.


DOUMIC Marie :

Title:Coming soon

Abstract: Coming soon


SALMON Stéphanie

Title: Numerical Model of Cerebral biological Flows using Real Biomedical Acquisitions,

Abstract: For several years, we have been studying the interactions between blood and cerebrospinal fluid (CSF) in the brain.
CSF is present throughout the central nervous system (brain and spinal cord) and provides mechanical protection for the brain and buoyancy, regulates intracranial pressure, and facilitates the elimination of toxins via the glymphatic system. We are developing several numerical models of varying dimensions and complexity to better understand these interactions and the dysregulation during the aging process.

These numerical simulations rely not only on medical images for geometric reconstruction,but also on clinical expertise and data measured bymagnetic resonance imaging (MRI). Indeed, MRI :allows us to quantify dynamic blood flow in intracranial vessels and CSF pulsatility.
We analyze cohorts of young and older subjects, as well as patients with intracranial hypertension for whom we also have access to pressure measurements.