No abstract

Qualitative modeling is a generic term that involves explicit and qualitative representations of the physical world. It can extend the realm of pure mathematical modeling in the sense that qualitative descriptions can, on one hand, simulate complex physical systems and processes and, on the other, produce linguistic descriptions and summaries of simulated system behavior. These summaries should be an essential element of the human/machine interface if truly interactive computational environments are to be developed. In the context of cardiac arrhythmias, a thorough understanding of the underlying processes that lead to the different pathological states is a first step toward optimizing diagnosis and therapy. The CARDIOLAB project is dedicated to cardiology and is aimed at providing a theoretical framework composed of computational models of different grain size and based on different formalisms. One of the intended roles of the framework is to assist researchers, clinicians, and pharmacologists in their quest for a better understanding of rhythmic disorders and ischemic events. In this paper, we present the first element of the framework. It is a cardiac simulator conceptualized in terms of a research field known as qualitative physics. As a simulator, the model's role is to produce fairly detailed descriptions, at different levels of abstraction, of cardiac electrical events when initial tissue-state conditions are given. A crude simulated ECG is also produced as a visual aid. At the end of each simulation session, and upon user request, the system can memorize the initial conditions and the descriptions into an arrhythmia knowledge base. As such, the model can be used as an interactive tool, to grossly delineate the regions in parameter space that correspond to causing or predisposing states leading to specific rhythmic disorders. More refined analysis can thereafter be performed using finer-grained models, the initial conditions of which will have been suggested by the qualitative model.