ABSTRACT
The problem of system identification in physiology derives its importance from the need to acquire quantitative models of physiologic function (from the subcellular to the organ system level) by use of experimental observations (data). Quantitative models can be viewed as summaries of experimental observations that allow scientific inference and organize our knowledge regarding the functional properties of physiologic systems. Selection of the proper (mathematical or computational) form of the model is based on existing knowledge of the system’s functional organization. System identification is the process by which the system model is determined from data. This modeling and identification problem is rather challenging in the general use of a physiologic system, where insufficient knowledge about the internal workings of the system or its usually confounding complexity prevents the development of an explicit model. The models may assume diverse forms (requiring equally diverse approaches) depending on the specific characteristics of the physiologic system (e.g., static/dynamic, linear/nonlinear) and the prevailing experimental conditions (e.g., noise contamination of data, limitations on experimental duration). This chapter will not address the general modeling issue, but rather it will concentrate on specific methods and tools that can be employed in order to accomplish the system identification task in most cases encountered in practice. Because of space limitations, the treatment of these system identification methods will be consistent with the style of a review article providing overall perspective and guidance while deferring details to cited references.
