Fast and robust parameter estimation with uncertainty quantification for the cardiac function

This repository contains the code accompanying the paper [1]. We employ an artificial neural network-based reduced-order model for cardiac electromechanics coupled with a physics-based 0D closed-loop blood circulation model [2] to perform fast parameter identification with inverse uncertainty quantification via Maximum a Posteriori estimation and Hamiltonian Monte Carlo by using a single core standard computer.

Installation

1. Install a conda environment containing all the required packages:

conda create -n envcardioEM-MAP python=3.7.11 numpy=1.21.5 matplotlib=3.5.1 pandas=1.3.4 scipy=1.7.3 mpi4py=3.0.3
conda activate envcardioEM-MAP
conda install -c anaconda scikit-learn
pip install --upgrade "jax[cpu]"

2. Clone this repository by typing:

git clone https://github.com/MatteoSalvador/cardioEM-MAP.git

3. Remember to activate the conda environment envcardioEM-MAP by typing conda activate envcardioEM-MAP (in case it is not already active from the installation procedure at point 1).

4. Choose the test case ('LV', 'atria' 'all') to run in run_MAP_estimation_ANN.py, along with the amount of noise in the observations (noise_std) and the number of trials (n_trials).

5. Execute the Python script run_MAP_estimation_ANN.py.

Note that also forward numerical simulations can be performed by using the Python script run_circulation_ANN.py.

Authors (alphabetical order)

• Francesco Regazzoni (francesco.regazzoni@polimi.it)
• Matteo Salvador (matteo1.salvador@polimi.it)

References

[1] M. Salvador, F. Regazzoni, L. Dede', A. Quarteroni. Fast and robust parameter estimation with uncertainty quantification for the cardiac function. Computer Methods and Programs in Biomedicine (2023).

[2] F. Regazzoni, M. Salvador, L. Dede', A. Quarteroni. A Machine Learning method for real-time numerical simulations of cardiac electromechanics. Computer Methods in Applied Mechanics and Engineering (2022).