Model-based interpretation of the ECG: a methodology for temporal and spatial reasoning

Abstract

A new software architecture for automatic interpretation of the electrocardiogram is presented. Using the hypothesize-and-test paradigm, a semi-quantitative physiological model and production rule-based knowledge are combined to reason about time- and space-varying characteristics of complex heart rhythms. A prototype system implementing the methodology accepts a semi-quantitative description of the onset and morphology of the P waves and QRS complexes that are observed in the body-surface electrocardiogram. A beat-by-beat explanation of the origin and consequences of each wave is produced. The output is in the standard cardiology ladder diagram format. The current prototype can perform the full differential diagnosis of 2:1 atrioventricular (AV) block, and can handle correctly complex rhythms such as AV nodal reentrant tachycardia with either hidden or visible P waves, and varying degrees of AV block.