Digital Twins

We create computationally-efficient and clinically-relevant tools and methods to rapidly build personalized computational models of how neural engineering interacts with the nervous system. Our goal is to accurately predict patient outcomes and efficiently identify the most effective and safe treatments for everyone.

Building on the pioneering work of Sin and Coburn and the foundational principles of Hodgkin and Huxley, our models assess the impact of physical fields on neural structures through mathematical approximations. We then explore the parameter space to efficiently identify effective and safe treatment modalities. We roughly categorize the components as follows:

  1. 3D models: We create workflows to generate anatomically accurate 3D models of the human body. We define and identify essential anatomical structures for accurate modeling and standard-of-care patient data acquisition protocols. We develop parametric 3D modeling, image processing, and computer vision methods for 3D volume generation.
  2. Multiphysics simulations: We develop and integrate multiphysics solvers, typically based on the finite element method, into high-performance computing frameworks to accurately model the distribution and interaction of external physical stimuli generated by medical treatments and devices in the human body.
  3. Biophysical neural models: We develop methods to accurately calculate physiological responses of neural structures and tissues to external stimuli. This involves creating, identifying, and validating a range of mathematical models of various physiological structures, balancing computational efficiency with biophysical realism.
  4. Optimization: We develop optimization routines to efficiently explore high-dimensional, potentially correlated parameter spaces in neural engineering treatments. These routines aim to identify Pareto-optimal solutions across multiple, potentially conflicting objectives to find a narrow set of effective and safe treatment parameters.
First Published on September 19, 2024
Last Updated on October 6, 2024 by Andreas Rowald