A hemodynamic abnormality such as hypovolemia or congestive heart failure frequently does not occur clinically as a discrete event, but rather may unfold in a sequence of clinical phases or “scenes”. This paper describes a prototype computer-based system being built to diagnose hemodynamic abnormalities by recognizing these scenes and the temporal relationships between them. Each scene is marked by certain physiologic findings which typically occur together. The paper therefore describes a paradigm for intelligent hemodynamic monitoring in which each diagnosis is defined as a sequence of clinical scenes, which we have named to reflect the predominant physiologic process involved; e.g., increased pericardial pressure, intravascular volume depletion, intravascular volume overload, vasoconstriction, vasodilation, and hypotension. These scenes are in turn defined by temporally related clusters of patient inputs.
展开▼
