National Library of Medicine

ion is the elimination of system components that are nonessential to the model and selection of those components that enable the model to best convey the intent of its designer. In computational modeling, the abstraction process may also be called reduction, variable elimination, or feature selection. For example, a patient may be associated with hundreds of variables but usually only a small portion of them is needed to model a patient in the context of the underlying illness. Aggregation is about the level of detail or granularity that is appropriate to the modeling task. In biomedical modeling, for instance, questions such as “should we consider a single cell, a cluster of cells with the same type, a tissue, an organ, a physiological system, an organism, or a population of organisms?” must be answered before constructing a functional model. In computational modeling, given a variable, the level of aggregation determines the range of the variable. Idealization is a set of assumptions on the system that are known to be slightly inaccurate or that do not always hold. Assuming our planet is geometrically a sphere (instead of an ellipsoid) is a typical idealization that is frequently made in our conversations. Assumptions about random variables, such as they are independent and identically distributed, or about time series, such as they are stationary, are typical examples of idealization in statistics. In the context of computational models, the necessity of simplification techniques can be analyzed in three dimensions: 1. Cognitive load. Simplification techniques may alleviate cognitive load of the designer and may prevent some complexity related errors. By yielding clarity in the 7 The sphere model of earth obviously abstracts out anything below the surface and aggregates all types of terrain structures into a uniformly smooth surface. SA MA stimulus