Computational models use mathematics, statistics, physics, and computer science to study the mechanism and behavior of complex systems. Also known as mathematical models, they contain numerous variables that characterize the system being studied. By adjusting these variables, a computer simulation can be performed to observe how changing these variables affects the outcomes predicted by the model.
Computer simulations can be used to quickly perform virtual experiments at a much lower cost than animal or human subject experiments. A wider variety of parameters can be evaluated, and confounding factors are easily controlled. Confounding factors are additional variables that can influence the outcome of a study if the researcher is unaware of or fails to account for them in the analysis. Computer simulation also enables possible outcomes to be evaluated that would not satisfy ethical boards for animal research (Institutional Animal Care and Use Committee – IACUC or ACUC) or human subject research (Institutional Review Board – IRB).
Computer models can be relatively simple or highly complex, and there are benefits to both. Simple models have fewer variables, and as a result, it is easier to understand the underlying relationships that govern system behavior. This modeling approach tends to be preferred by physicists. Highly complex models have many variables and often complex interactions between variables. Because these models account for so many variables, they can predict behavior of a system that would be difficult to predict intellectually without the use of modeling software.
Although models are useful in predicting system behavior, they are only an approximation of the real system. Even highly complex models with hundreds of variables are far from the infinite degrees of freedom in an actual system (even a simple one). Because of this, physical experiments should also be performed to verify accuracy of computational models.
