Abstract Models That Predict or Extend Animal Bioassays

Abstract

For about 30 years computer-based mathematical models have been designed to aid the analysis and understanding of physiological mechanisms. These models are a natural extension of conceptual reasoning and support the research process by explicitly detailing the hypothesized connections and causality, the logical structure of the real system under study. Unfortunately, in contrast to engineering, abstract models in physiology are inherently less credible owing to lack of strict verisimilitude in the phenomological or even in the mechanistic simulations. Consequently, modeling in physiology has been less satisfactory than in, say, engineering or physics. However, there are now many examples of satisfactory models that have proved to be both valid and exceedingly useful. It is the purpose of this paper to describe the characteristics of a useful model, to explain what is meant by validity, and to show by example how judicious interaction between conceptual modeling and the experimental laboratory can become a symbiotic process of power and efficiency. In particular, the model can not only be used to interpolate and extrapolate the range of parameters obtained from few well chosen experiments, but can also aid the design of the experiment by formulating better questions. In effect the experiment should, and in many cases can now, be run on the computer, with animals used only as corroborating evidence. Obviously there are still many limitations and circumstances in which this is not yet possible, but the trend is clearly marked by current successes. There is good reason to believe that valid mathematical models can be designed for partial fulfillment of the requirements for animal testing.