Stochastic Modeling of Biological Systems – Ranking the Model Parameters of the Human Vocal Folds

Computational models of biological systems are becoming more and more common in medical research areas. Evidence of this can be found by examining the number of articles containing the term “finite element” in the expansive National Institutes of Health (NIH) digital research archive PubMed. Between the years 1985 and 2005, the number of research studies that referenced the finite element method increased 29-fold from 206 in 1985 to nearly 6000 in 2005. Numerical modeling of biological systems allows the execution of “computational experiments” that are impossible to duplicate in reality, and provides many unique insights for biological research questions. However, unlike typical engineering systems, biological systems are characterized by extremely high levels of uncertainty. In biological modeling, many geometric and material parameters which are required for the solution of finite element models are either poorly understood, or have often never been measured experimentally. One way to circumvent this challenge is to use methods of stochastic modeling to create a “sample set” of numerical models in much the same way that medical researchers use a sample when conducting traditional medical research. This paper presents the use of COMSOL Multiphysics to generate a random sample of computational vocal fold models. Statistical analysis of these models was then used to rank the model parameters according to their influence on the model behavior, thus reducing the number of parameters that must be considered in future experimental and computational research studies.