Knowledge-based systems and neural networks for clinical decision making

This paper presents two knowledge-based systems (KBS) and an artificial neural network (ANN) system for clinical decision-making in electrocardiogram (ECG) signal interpretation. Among these systems, a KBS contains “shallow” knowledge in declarative forms and employs fuzzy set theory to deal with vagueness in the encoded knowledge and imprecise ECG measurements. The other KBS uses “deep” knowledge encoded in a qualitative simulation model for ECG simulation and interpretation. An experimental ANN was constructed to test its usefulness for ECG interpretation. Preliminary results show that each system has its own usefulness in ECG interpretation and simulation. Brief comparisons are also given in the paper.