IMAG Discussion Topic

[Trent Guess]

Following is the discussion topic for the April IMAG meeting. Please discuss the discussion topic. I believe that we should have some direction and a solid foundation on the topic prior to the meeting.

"Determination of computational priorities and challenges related to multiscale modeling in computational biomechanics."

[Ahmet Erdemir]

Computational Priorities:

1. Need for simulation tools designed for the biological system: Using an engineering simulation software such as ABAQUS (finite element solver) does not necessarily provide optimal convergence when simulating biomechanical phenomena.

2. Software to investigate multiple domains, e.g. fluid-solid coupling, exist. However, simulation tools that allow multiple scales or glue different software platforms do not exist (at least I am not aware of).

3. Computational power is developing at a high-pace allowing simulations run faster but model development capacity does not develop that fast. Generating meshes are cumbersome, linking different scales may be tedious, etc.

Basically, I am more concerned about model development cost than the simulation cost.

Challenges:

1. What to focus? Applied science by investigating a clinical problem or computational research by developing algorithms.

2. Literature has a lot of models as well as input data for almost any form of modeling. Let's say that I want to use a cell level model to bridge mechanical loading at skeletal level to cell death, what kind of cell level model should I use given that I have very limited knowledge to judge what is good in literature. Of course, one can say that I need a collaborator. How would I find one that I believe suitable for my application.

3. Validation. This is already challenging without spanning multiple scales. How can we deal with this at with multiscale modeling? Validate macroscopic and microscopic models, separately and hope they work together well.

4. Experiments. Obviously, necessary for model input and validation. What are standard experiments used in multiple disciplines for these purposes?

5. Translation of results. Again, for investigator's of different disciplines, this topic becomes important.

[Trent Guess]

Another computational priority

Repositories. Specifically, a repository that contains supporting information for development of multi-scale computational models would be useful.

For computational biomechanics, this repository might contain:

- geometries body, organ, tissue…
- material properties various scales
- experimental data various scales
- anthropometric data

Obviously there are several concerns with a repository, the least of which may be the large variety of formats that this information would be in.

To maintain the integrity of the repository, repository entries should be supported by peer reviewed journal articles.

For the repository to work, it would need to be sufficiently funded and maintained.

Visual Human Project is essentially a repository that researchers have found useful. Many musculoskeletal projects utilize these geometries. SimTK also serves as a repository.

One final thought. As there may not be much incentive for researchers to submit to a repository, the incentive may come from NIH and NSF. Ie. a stipulation that the result of certain funded projects should be posted to a repository, similar to the model sharing criteria.

[Jeff Bischoff]

These are sensible topics, and would be interesting to discuss. Another approach, just to get the discussion rolling at the IMAG meeting, is to frame our discussion around an article of some support, from which we can branch off into areas of interest. Alternatively or in addition, maybe the group could document the discussion using the paper opportunity in IEEE/EMBS.