What is PAGE?

We represent a community with a shared interest in data analysis using the population approach.


2001
   Basel, Switzerland

Integrated system models to understand disease status, progression and therapy: case studies

Paolo Vicini

University of Washington, United States, 98195-2255, Seattle, Box 352255

Integrated systems modeling can be defined as the application of mathematics and statistics to understand the transient behavior of biological systems. The model organizes the components characterizing the system and the transport or transformation of substances among these components in a testable mathematical framework. Substances include those already present in the system - endogenous compounds - and those foreign to the system - xenobiotics such as drugs. In either case, one seeks a mathematical description of the system that is consistent with known physiology and pharmacology, and is compatible with all available data. The models are then subject to identifiability and validity tests and used to estimate unknown parameters of interest, to make predictions about system behavior, to simulate previously unobserved behavior in response to a perturbation, and to aid in experimental design.

In a broader context, however, the focus of integrated system models lies in understanding the mechanisms of disease and quantifying the effect of therapeutic interventions. Data that can be analyzed within this framework characterize the time dependent status or progression of a disease. We will show that, with appropriate experimental design and data, the combination of kinetic, dynamic and disease status modeling can aid enormously in understanding: changes in disease severity as a function of time and in response to different therapeutic interventions; the kinetics and dynamics of a pharmacologic response in the clinically relevant context of a changing disease status; sources of variability in disease status and therapeutic response; therapy regimen individualization; estimation and prediction of clinical outcomes in pharmacoeconomic studies. Examples of data suitable for such analyses include various biomarkers or surrogate endpoints for clinical response: continuous variables such as those arising from imaging data, or discrete variables such as categorical assessment tests. The actual model formulation is a multidisciplinary task and comes from a basic understanding of the disease and the modeler's interactions with collaborators expert in diagnosing and treating the disease. In this talk, we will review successful examples of such interactions in several fields of the modern biomedical literature.



Top