When completed, the model will have a web-based user interface that allows immunologists and other researchers to interact with the model, gain insights into the early stages of the infection and test potential alternative, host centric interventions.

Neural representations of for instance task-relevant information are often distributed across multiple brain areas. These representations could have independent dynamics, or could influence each other. Our tool uses latent dynamical system models to assess the degree of influence of dynamics in one brain area on another by correlating the sub-single trial fluctuations of brain areas.

Tissue growth and remodeling (G&R) has been modeled in a continuum mechanics framework with an approach similar to plasticity named 'finite growth model'. Computationally, G&R has been simulated with custom finite element implementations. However, these computational implementations are computationally expensive and require an expert modeler to setup for any new set of parameters, boundary or initial conditions. Thus, current models capture the overall trends of G&R in different applications but are difficult to calibrate, cannot be evaluated easily, and ignore mechanical and biological uncertainty. To address these limitations, we leverage machine learning tools to replace the finite element simulations with an inexpensive surrogate with quantified epistemic uncertainty. Specifically, we look at skin growth during tissue expansion and propose a multi-fidelity Gaussian process surrogate to replace the finite element solver. The methodology can be extended to other applications of G&R. We have published one article and the code is available through Bitbucket (link below).