Section: New Results

Numerical methods for cardiac electrophysiology

Participants : Muriel Boulakia, Jean-Frédéric Gerbeau, Damiano Lombardi, Fabien Raphel.

In [19] a method to assess the variability of phenomena described by PDEs is proposed. In particular, the probability density distribution of the parameters of a model is estimated, in such a way that the statistics of the model output match the observed ones. The investigated approach is based on a differential entropy regularised moment matching.

In [25] we investigated how, by a semi-empirical design of composite biomarkers, the classification of the action of a drug on the electrical activity of a cell can be improved. The data used are measured with a Micro-Electrodes-Array.

In [33] a method is investigated, to design composite biomarkers by exploiting a database of in silico experiments. In particular, a dictionary approach is proposed. The composite biomarker is expressed as a linear combination of linear and non-linear forms applied to the observable. The coefficients of the combination are determined by solving a ${\ell }^1$ regularised optimisation problem.